Top 10 Strategic Technology Trends For 2020

Top 10 Strategic Technology Trends for 2020
Published: 21 October 2019 ID: G00432920
Analyst(s): David Cearley, Nick Jones, David Smith, Brian Burke, Arun Chandrasekaran, CK Lu
Strategic technology trends have the potential both to create opportunity

and to drive significant disruption. Enterprise architecture and technology
innovation leaders must evaluate these top trends to determine how
combinations of trends can power their innovation strategies.
Key Findings
■ Strategic technology trends have significant potential to create and respond to disruption and to
power both transformation and optimization initiatives.
■ Artificial intelligence (AI) is a foundational catalyst for advanced process automation and human
augmentation and engagement.
■ Physical environments including factories, offices and cities will become “smart spaces” within
which people will interact through multiple touchpoints and sensory channels for an increasingly
ambient experience.
■ Dealing with privacy, digital ethics and security challenges generated by AI, the Internet of
Things (IoT)/edge, and other evolving technologies will become critical to maintain trust and
avoid legal entanglements.
Recommendations
Enterprise architecture and technology innovation leaders must:
■ Center their innovation efforts on people and use tools such as personas, journey maps,
technology radars, and roadmaps to evaluate opportunities, challenges and time frames for
adoption.
■ Build an overarching view across functional and process silos and exploit a complementary set
of tools including RPA, iBPMS, DTO, application development, and AI domains that guide how
the tools are used and the systems they create are integrated.
■ Embrace multiexperience and implement development platforms and design principles to
support conversational, immersive and increasingly ambient experiences.
■ Establish governance principles, policies, best practices and technology architectures to
increase transparency and trust regarding data and the use of AI.
Table of Contents
Strategic Planning Assumptions............................................................................................................. 3

Analysis.................................................................................................................................................. 4
People-Centric Smart Spaces Build on the Intelligent Digital Mesh................................................... 6
Trend No. 1: Hyperautomation........................................................................................................10
RPA and iBPMS Are Key Components of Hyperautomation......................................................10
The Digital Twin of an Organization........................................................................................... 11
Machine Learning and NLP Explode the Range of Hyperautomation Possibilities......................12
Trend No. 2: Multiexperience.......................................................................................................... 13
Immersive Experiences............................................................................................................. 14
The Language Factor................................................................................................................15
The Multiexperience Development Platform.............................................................................. 16
Trend No. 3: Democratization......................................................................................................... 17
Democratization of Application Development............................................................................ 18
Low-Code, No-Code and the Citizen Developer....................................................................... 19
Augmented Analytics and the Citizen Data Scientist................................................................. 19
Dealing With Shadow AI........................................................................................................... 20
Trend No. 4: Human Augmentation.................................................................................................21
Cognitive and Physical Augmentation....................................................................................... 22
Cultural and Ethical Aspects of Human Augmentation.............................................................. 23
Trend No. 5: Transparency and Traceability.....................................................................................25
Explainable and Ethical AI and Algorithms.................................................................................25
Data Privacy, Ownership and Control........................................................................................26
Best Practices and Regulations Are Emerging.......................................................................... 28
Trend No. 6: The Empowered Edge................................................................................................29
Data, Analytics and AI at the Edge............................................................................................30
Communicating to the Edge — The Role of 5G........................................................................ 31
Digital Twin of Things at the Edge............................................................................................. 32
Trend No. 7: Distributed Cloud....................................................................................................... 33
Distributed Cloud Delivers on the Hybrid Cloud Promise........................................................... 33
Getting to Distributed Cloud..................................................................................................... 34
The Edge Effect........................................................................................................................ 35
Page 2 of 52 Gartner, Inc. | G00432920

Trend No. 8: Autonomous Things................................................................................................... 36
Common Technology Capabilities............................................................................................. 36
Combining Autonomous Capabilities and Human Control......................................................... 37
Autonomous, Intelligent and Collaborative................................................................................ 38
Trend No. 9: Practical Blockchain................................................................................................... 39
Blockchain Will Be Scalable by 2023........................................................................................ 40
Blockchain Use Cases..............................................................................................................41
Trend No. 10: AI Security................................................................................................................43
Protecting AI-Powered Systems............................................................................................... 43
Leveraging AI to Enhance Cybersecurity Defense..................................................................... 44
Anticipating Nefarious Use of AI by Attackers........................................................................... 45
List of Figures
Figure 1. Top 10 Strategic Technology Trends for 2020...........................................................................5

Figure 2. People-Centric Smart Spaces..................................................................................................7
Strategic Planning Assumptions

By 2022, 70% of enterprises will be experimenting with immersive technologies for consumer and
enterprise use, and 25% will have deployed them to production.
By 2022, 35% of large businesses in the training and simulation industry will evaluate and adopt
immersive solutions, up from less than 1% in 2019.
By 2021, at least one-third of enterprises will have deployed a multiexperience development

platform to support mobile, web, conversational and augmented reality development.
By 2024 75% of large enterprises will be using at least four low-code development tools for both IT
application development and citizen development initiatives.
By 2022, at least 40% of new application development projects will have artificial intelligence co-
developers on the team.
By 2021, automation of data science tasks will enable citizen data scientists to produce a higher
volume of advanced analysis than specialized data scientists.
By 2025, a scarcity of data scientists will no longer hinder the adoption of data science and
machine learning in organizations.
Gartner, Inc. | G00432920 Page 3 of 52

By 2022, 30% of organizations using AI for decision making will contend with shadow AI as the
biggest risk to effective and ethical decisions.
Through 2023, 30% of IT organizations will extend BYOD policies with “bring your own
enhancement” (BYOE) to address augmented humans in the workforce.
By 2020, we expect that companies that are digitally trustworthy will generate 20% more online
profit than those that aren’t.
By 2020, we expect that 4% of network-based mobile communications service providers (CSPs)

globally will launch the 5G network commercially.
By 2024, most cloud service platforms will provide at least some services that execute at the point
of need.
By 2023, blockchain will be scalable technically, and will support trusted private transactions with
the necessary data confidentiality.
Through 2022, over 75% of data governance initiatives will not adequately consider AI’s potential
security risks and their implications, resulting in quantifiable financial loss.
Through 2022, 30% of all AI cyberattacks will leverage training-data poisoning, AI model theft or
adversarial samples to attack AI-powered systems.
Analysis
Gartner’s annual top 10 strategic technology trends (see Figure 1) highlight trends that enterprises
need to consider as part of their five-year strategic technology planning process. These trends have
a profound impact on people and the spaces they inhabit. Strategic trends have broad impact
across industries and geographies, with significant potential for disruption. In many cases, the
trends aren’t yet widely recognized or the conventional wisdom regarding the trends is shifting (see
Note 1). Through 2025, technologies related to these trends will experience significant changes,
cross critical tipping points and reach new levels of maturity that expand and enable repeatable use
cases and reduce risk. Technology innovation leaders must examine the business impact of our top
10 strategic technology trends and seize the opportunities to enhance existing, or create new,
processes, products and business models. Prepare for the impact of these trends — they will
transform industries and your business.

Figure 1. Top 10 Strategic Technology Trends for 2020
The top 10 strategic technology trends are a critical ingredient in driving a continuous innovation
1
process as part of a ContinuousNext strategy. Technology innovation leaders must adopt a mindset
and new practices that accept and embrace perpetual change. That change may be incremental or
radical, and may be applied to existing or new business models and technologies.
Technology is changing people’s lives, enabling the ongoing digitalization of business and driving
organizations to continually refresh their business models. In this long march to digitalization,
technology is amplifying continuous change at an ever-increasing velocity. Organizations need to
consider how these strategic trends can be applied across two continuous and complementary
cycles:
■ Continuous operations exploit technology to run the business today, modernize it and improve
efficiency, and incrementally grow the business. Existing business models and technology
environments set the stage upon which innovation opportunities are explored and will ultimately
influence the cost, risk and success of implementing and scaling innovation efforts.
■ Continuous innovation exploits technology to transform the business and either create or
respond to disruptions affecting the business. This innovation cycle looks at more radical
changes to business models and supporting technologies as well as new business models and
technologies that extend the current environment.
Trends and technologies do not exist in isolation. They build on and reinforce one another to create
the digital world. Combinatorial innovation explores the way trends combine to build this greater

whole. Individual trends and related technologies are combining to begin realizing the overall vision
2
embodied in the intelligent digital mesh. For example, AI in the form of machine learning (ML) is
combining with hyperautomation and edge computing to enable highly integrated smart buildings
and city spaces. In turn, this enables further democratization of the use of technology as business
users and customers self-serve to meet their individual needs. CIOs and IT leaders are challenged
with a number of “and dilemmas” such as delivering existing services and innovating new services.
You can turn “and dilemmas” into “and opportunities” by applying “and thinking,” where the
combined effect of multiple trends coalescing is to produce new opportunities and drive new
disruption. Combinatorial innovation is a hallmark of our top 10 strategic technology trends for
2020.
People-Centric Smart Spaces Build on the Intelligent Digital Mesh

The future will be characterized by smart devices delivering increasingly insightful digital services
everywhere. We call this the “intelligent digital mesh.” It can be described as:
■ The intelligent theme explores how AI — with a particular emphasis on machine learning — is
seeping into virtually every existing technology and creating entirely new technology categories.
The exploitation of AI will be a major battleground for technology providers through 2022. Using
AI for well-scoped and targeted purposes delivers more flexible, insightful and increasingly
autonomous systems.
■ The digital theme focuses on blending the digital and physical worlds to create a natural and
immersive digitally enhanced experience. As the amount of data that things produce increases
exponentially, compute power shifts to the edge to process stream data and send summary
data to central systems. Digital trends, along with opportunities enabled by AI, are driving the
next generation of digital business and the creation of digital business ecosystems. In this
intelligent digital world, massive amounts of rich and varied data are generated, thus
necessitating a greater focus on digital ethics and privacy and on the transparency and
traceability that they require.
■ The mesh theme refers to exploiting connections between an expanding set of people and
businesses — as well as devices, content and services — to deliver digital business outcomes.
The mesh demands new capabilities that reduce friction, provide in-depth security and respond
to events across these connections.
Intelligent digital mesh has been a consistent theme of Gartner’s strategic technology trends for the
past few years, and it will continue as an important underlying force over the next five years.
However, this theme focuses on the set of technical characteristics of the trend. It is also important
to put the trends in the context of the people and organizations that will be impacted. Rather than
building a technology stack and then exploring the potential applications, organizations must
consider the business and human context first. We call this “people-centric smart spaces” and this
structure is used to organize and evaluate the primary impact of the strategic trends for 2020 (see
Figure 2).

Figure 2. People-Centric Smart Spaces
By putting people at the center of your technology strategy you are highlighting one of the most
important aspects of technology — how it impacts your customers, employees, business partners,
3
society or other key constituencies. A people-centric approach should start with understanding
these key target constituencies and the journey they undertake to interact with or support your
organization. This is the first step to understanding how and where you will apply strategic
technology trends to drive desired business outcomes:
■ Personas: The persona is a useful tool to describes a target individual or group. The persona
encapsulates a set of motivations, preferences, biases, needs, wants, desires and other
characteristics that can be used as a backdrop to evaluate how applications of technology
might impact that group. Personas have been used for many years and have gained broadest
adoption in design and marketing areas, where understanding the motivations of a target
4
audience are particularly important. The persona sets the context for evaluating the potential
impact on people and the resulting business outcome. Personas can be used to anticipate the
5
valuable business moments that emerge as people traverse technology-enabled smart spaces.

Persona-based analysis is a powerful tool that helps leaders diagnose and take action against
digital-business-disruption opportunities. Enterprise architecture and technology innovation
leaders can help business and IT leaders to consider the human side of digital business strategy
decisions with personas.
■ Journey Maps: A second useful tool is the defining of “journey maps.” A journey map is a
model that shows the stages that target personas go through to accomplish a task or complete
a process. Customer journey maps diagram the stages a customer might go through to buy
products, access customer service, or complain about a company on social networks. One can
also consider internal journey maps that diagram the stages employees go through in
onboarding or in complying with a regulatory requirement. Journey maps that look at how
multiple constituencies interact around a process are even more powerful. For example, a
journey map for a customer purchase might consider not only the customer view, but also that
of a salesperson or a fulfillment group. Journey maps provide even more concrete context for
technology-driven innovation. Technology innovation professionals should consider the pain
points, inefficiencies, gaps, and opportunities to delight and create new digital business
moments for all the relevant constituents.
The concept of smart spaces builds on the people-centric notion. People exist in spaces such as
their homes, their cars, an office building, a conference room, a hospital or a city. A smart space is a
physical environment in which humans and technology-enabled systems interact in increasingly
open, connected, coordinated and intelligent ecosystems. Multiple elements — including people,
processes, services and things — come together in a smart space to create a more immersive,
interactive and automated experience for a target set of personas. Journey maps should consider
not only the motivations of relevant personas and the desired business outcomes, but also the
spaces that people will traverse as part of their interactions in the digital world.
People-centric smart spaces represent the core target for applying technology trends. Surrounding
this focus, technology innovation leaders should consider a concentric ring of trends that progress
from those that directly touch the human to those that are more hidden with a derivative impact.
Thinking of the technology environment from the center out, proceeding from people and ultimately
focusing on the back-end infrastructure, is a subtle but useful shift from the traditional bottom-up
view of a technology stack. Driving architecture from the center out ensures requirements at each
level are driven by business outcomes and the architecture is inherently more adaptable:
■ People interact with others and with the digital spaces through edge devices and multiple user
experiences (interfaces) across these edge devices as a natural part of their everyday lives.
■ Applications, services, and ecosystems deliver value to people in those spaces through edge
devices and experiences.
■ Development tools, digital platforms, data, AI/ML services, integration and related technologies
are used to create the applications and services that deliver value through edge devices. These
tools provide a dynamic, flexible and modular foundation to create the applications, services
and ecosystems.

■ Infrastructure, operations, networking and security are the foundation upon which the
technology-enabled world is built. It must deliver value at each layer as it ultimately supports
people.
Our top 10 strategic technology trends for 2020 are organized into people-centric and smart spaces
categories. This is a loose organization intended to convey where the primary impact and
manifestation of the trend will be seen. However, virtually all of the trends will have an impact on
both the people and smart spaces concepts.
■ People-Centric:
■ Hyperautomation deals with the application of advanced technologies including AI and
machine learning to increasingly automate processes and augment humans.
■ Multiexperience deals with the way that people perceive, interact and control the digital
world across a wide range of devices and sensory touchpoints.
■ Democratization explores how to create a simplified model for people to consume digital
systems and tap into automated expertise beyond their training or experience.
■ Human augmentation explores how humans are physically and cognitively augmented by
these systems.
■ Transparency and traceability focuses on the data privacy and digital ethics challenges
and the application of design, operational principles, and technologies to increase
transparency and traceability to enhance trust.
■ Smart Spaces:
■ Empowered edge emphasizes how the spaces around us are increasingly populated by
sensors and devices that connect people to one another and to digital services.
■ Distributed cloud examines a major evolution in cloud computing where the applications,
platforms, tools, security, management and other services are physically shifting from a
centralized data center model to one in which the services are distributed and delivered at
the point of need. The point of need can extend into customer data centers or all the way to
the edge devices.
■ Autonomous things explores how physical things in the spaces around people are
enhanced with greater capabilities to perceive, interact, move, and manipulate these spaces
with various levels of human guidance, autonomy and collaboration.
■ Practical blockchain focuses on how blockchain can be leveraged in practical enterprise
use cases that are expanding over the next three to five years.
■ AI security deals with the reality of securing the AI-powered systems that are behind the
people-centric trends.

Trend No. 1: Hyperautomation
Automation refers to the use of technology to facilitate or perform tasks that originally required
some form of human judgment or action. The term “tasks” refers not only to tasks and activities in
the execution, working or operational environment, but it also encompasses tasks in thinking,
discovering and designing these automations themselves.
Hyperautomation refers to the combination of multiple machine learning, packaged software and
automation tools to deliver work. The propensity to use particular types of automation will be highly
dependent on the organization’s existing IT architecture and business practices. Hyperautomation
refers not only to the breadth of the pallet of tools, but also to all the steps of automation itself
(discover, analyze, design, automate, measure, monitor, reassess).
Hyperautomation is an unavoidable market state in which organizations must rapidly identify and
automate all possible business processes with several implications:
■ The scope of automation changes. The automation focus now spans automating individual,
discrete tasks and transactions based on static and rigid rules, to automating more and more
knowledge work. In turn, those levels of automation enable enhanced and more dynamic
experiences and better business outcomes.
■ A range of tools will be used to manage work and coordinate resources. Increasingly,
organizations will use an evolving set of technologies to support an ever-expanding business
scope. The tools include task and process automation, decision management, and packaged
software — all of which will incorporate more and more machine learning technologies.
■ Architecting for agility is required. This means organizations need the ability to reconfigure
operations and supporting processes in response to evolving needs and competitive threats in
the market. A hyperautomated future state can only be achieved through hyperagile working
practices and tools.
■ Workforce engagement is needed to reinvent how employees deliver value. Without
engaging employees to digitally transform their operations, the organization is destined to gain
only incremental benefits. This means overcoming the challenges associated with silos and the
way the organization allocates resources and integrates the capabilities of its partners and
suppliers.
RPA and iBPMS Are Key Components of Hyperautomation

Hyperautomation requires selection of the right tools and technologies for the challenge at hand.
Understanding the range of automation mechanisms, how they relate to one another, and how they
are combined and coordinated, is a major focus for hyperautomation. This is complicated because
there are currently many multiple, overlapping and yet ultimately complementary technologies
including:
■ Robotic process automation (RPA): RPA is a useful way to connect legacy systems that don’t
have APIs with more modern systems (see Note 2). It will move structured data from system A
to system B better than people and addresses integration challenges with legacy systems. The

scope of these processes is typically a short-lived task associated with moving that data. RPA
tools can also augment knowledge workers undertaking day-to-day work, removing mundane
and repetitive tasks. Tightly defined integration scripts structure and manipulate data, moving it
from one environment to the next. Because the integration is based on interacting with the
metadata that drives the screens of existing applications, these tools are usually more
accessible to business end users.
■ Intelligent business process management suites (iBPMSs). Beyond RPA, iBPMSs manage
long-running processes. An intelligent business process management suite is an integrated set
of technologies that coordinates people, machines and things. An iBPMS relies on models of
processes and rules to drive a user interface and manage the context of many work items
based on those models. Integration with external systems is usually achieved via robust APIs.
Alongside the processes, strong decision models can simplify the environment and provide a
natural integration point for advanced analytics and machine learning. iBPMS software supports
the full life cycle of business processes and decisions: discovery, analysis, design,
implementation, execution, monitoring and continuous optimization. An iBPMS enables citizen
developers — most commonly business analysts — and professional developers to collaborate
on iterative development and improvement of process and decision models.
These technologies are highly complementary, and Gartner increasingly sees them deployed side by
side.
iBPMS platforms can choreograph complex styles of work — for example, adaptive case
6
management or processes driven by complex events. This is increasingly important, particularly in
the context of digitalized processes that coordinate the behaviors of people, processes and the
“things” that are part of the Internet of Things. The rapidly changing operational context in a
digitalized process requires actionable, advanced analytics to more-intelligently orchestrate
business processes across the virtual and physical worlds.
A key thing to recognize is that the organization is becoming more model-driven, and the ability to
manage the interconnected nature and complex versioning of these models is an important
competency. To derive the full benefits of hyperautomation, organizations need an overarching view
across their functional and process silos. Indeed, developing more and more sophisticated models
is akin to developing a digital twin of an organization (DTO).
The Digital Twin of an Organization

A digital twin of an organization visualizes the interdependence between functions, processes and
key performance indicators (KPIs). A DTO is a dynamic set of software models of a part of an
organization. It relies on operational and/or other data to understand and provide continuous
intelligence on how an organization operationalizes its business model connected directly to its
current state and deployed resources. Critically, a DTO responds to changes in how expected
customer value is delivered.
A DTO draws from a real-world environment with real people and machines working together to
generate continuous intelligence about what is happening throughout the organization. Effectively, a

DTO provides a contextual framework for business process and decision models. It helps capture
where enterprise value links to the different parts of an organization and how its business processes
impact value creation. As such, the DTO becomes an important element for hyperautomation. A
DTO allows users to model and explore scenarios, choose one, and then make it real in the physical
7
world.
Machine Learning and NLP Explode the Range of Hyperautomation Possibilities

AI techniques in various forms, including machine learning and natural language processing (NLP),
have rapidly expanded hyperautomation possibilities. Adding the ability to interpret human speech
quickly at runtime, identify patterns in documents or data, and/or dynamically optimize business
outcomes dramatically alters the range of automation possibilities. Indeed, when combined with
RPA- and iBPMS-related products, they are starting to affect many industries and enable the
automation of what was once deemed the exclusive domain of knowledge workers. But, rather than
replacing those workers, AI technologies are mostly augmenting their ability to deliver value. These
AI technologies have created a mini arms race as vendors attempt to leapfrog each other. AI
technologies have:
■ Improved machine vision functionality in most RPA tools. Indeed, machine learning has
enabled a major step forward in RPA through a type of computer vision. For example, it can
recognize a “Submit” button and virtually press it, regardless of where it might appear on the
screen. This has been extended to identifying all the text on screen (just like optical character
recognition [OCR]). Taking this a step further, tools are emerging that can separate the labels on
an image from the dynamically populated text fields. This innovation then enables the RPA tool
to interact with image-based interfaces as though they were directly accessed applications.
■ Optimized business KPIs. An iBPMS or RPA tool can easily call a machine learning model or
NLP functionality directly from within the short-term task or longer-running business process.
Increasingly, machine learning and NLP are being directly embedded into iBPMS tools with
preintegrated functionality making it easier to do the associated data science (plug-and-play
machine learning) or call external services from the mega cloud vendors such as Amazon,
Google, IBM and Microsoft.
■ Appeared in a plethora of adjacent and supporting technologies. These include advanced
OCR and intelligent character recognition (ICR) to interpret handwriting. NLP is enabling more
and more self-service automation as customers interact directly with chatbots and virtual
personal assistants (VPAs).
■ Automated processes discovery. Machine learning is enabling vendors to discover working
practices and their different variants in the workplace. Task mining tools, sometimes referred to
as “process discovery,” help organizations achieve deep insight into their task flows to get the
8
microview of the task steps or activities that could then be automated by RPA or an iBPMS.
The addition of machine learning and NLP to RPA and iBPMS tools provides the ability to
industrialize the digital customer and employee experience by connecting those interactions directly
to automated back-office operations and supplier ecosystems. Moreover, this enables a
contextually aware, situationally adaptive approach — where the set and order of interactions

among participants are uniquely choreographed based on the goals of the business, its partners,
and its customers, and on operations intelligence that continuously updates and analyzes in real
time. An iBPMS can proactively personalize contextually aware interactions at scale while
supporting rapid transformation and/or improvement of the customer and employee journey.
Furthermore, AI supports an iBPMS in automating and orchestrating business processes that shape
themselves as they run. These processes can therefore be considered adaptive and intelligent,
executing the best, next action instead of the same repeatable sequence of actions.
Related Research:
■ “Navigate Optimal Routes for Process Automation With RPA, iBPMS and iPaaS”
■ “Market Guide for Technologies Supporting a DTO”
■ “Develop 3 Levels of Service for Your Center of Expertise to Scale DigitalOps and Robotic
Process Automation”
■ “Artificial Intelligence Trends: Process Augmentation”
■ “The State of RPA Implementation”
■ “Best Practices for Robotic Process Automation Success”
■ “Create a Digital Twin of Your Organization to Optimize Your Digital Transformation Program”
■ “Comparing Digital Process Automation Technologies Including RPA, BPM and Low-Code”
Trend No. 2: Multiexperience

Through 2028, the user experience will undergo a significant shift in how users perceive the digital
world and how they interact with it. Conversational platforms are changing the way in which people
interact with the digital world. Virtual reality (VR), augmented reality (AR) and mixed reality (MR) are
changing the way in which people perceive the digital world. This combined shift in both perception
and interaction models leads to the future multisensory and multitouchpoint experience.
The model will shift from one of technology-literate people to

one of people-literate technology. The burden of translating
intent will move from the user to the computer.
The ability to communicate with users across many human senses will provide a richer environment
for delivering nuanced information.
The “computer” in a multiexperience world is the environment around the user, including many
touchpoints and sensory inputs. The multitouchpoint aspect of the experience will connect people
across edge devices, including traditional computing devices, wearables, automobiles,
9
environmental sensors and consumer appliances. The multisensory aspect of the experience will
use all human senses as well as advanced computer senses (such as heat, humidity and radar) as

appropriate across this rich sea of devices. In the future, the very notion of “computer” will seem
like a quaint and antiquated idea as the spaces that we inhabit become multisensory and
multitouchpoint interfaces.
The long-term manifestation of multiexperience is also called the ambient experience. However, this
will happen only slowly through 2029 and beyond. Privacy concerns in particular may dampen the
enthusiasm and impact of adoption. On the technical front, long life cycles of many consumer
devices, and the complexity of having many creators developing elements independently, will be an
enormous barrier to seamless integration. Don’t expect automatic plug and play of devices,
applications and services. Instead, there will be proprietary ecosystems of devices in the near term.
Focus on targeted use of immersive experiences and conversational platforms for highly specific
scenarios as they evolve through 2024. These experiences and platforms will overlap with one
another, incorporating a full array of sensory input/output channels delivered in various device
scenarios. These may be targeted experiences on specific devices, but opportunities will grow for
more robust scenarios across multiple devices and sensory channels to support specific
environments (e.g., an ambient experience in a manufacturing plant). Complement these targeted
solutions with an evolving multiexperience development platform that incorporates more sensory
channels and more device targets well beyond the web and mobile targets of more traditional
development platforms.
Immersive Experiences
Immersive experience is constructed using a variety of techniques and software tools, including AR,
VR, MR, multichannel human-machine interface (HMI) and sensing technology. Immersive
experience is distinguished from other experience methods by its ability to simulate realistic
scenarios and environments that give users the visualization and the actionable information to
practice operations and to interact with virtual people or objects. These experiences include a wider
range of immersions from simple AR overlays on a smartphone to fully immersive virtual reality
environments.
VR provides a 3D environment that surrounds a user and responds to an individual’s actions in a

natural way. This may be through an immersive head-mounted display (HMD) that blocks the user’s
entire field of vision. HMDs enrich the immersive experience by blending the digital and the
physical. But the current-generation product has many constraints, including heavy power
consumption, clumsy designs, awkward UI, latency and limited field of view.
Smartphones can also be an effective platform for mobile VR and AR. Accessories exist to turn
smartphones into HMDs. But an HMD configuration is not vital to experience AR — AR can
combine digital overlays on a real-world video experience. The smartphone’s screen becomes a
“magic window” that displays graphics overlaid on top of real-world things. AR superimposes
contextual information that blends augmented data on top of real-world objects (such as hidden
wiring superimposed on an image of a wall). Although this approach has limitations compared with
HMD-based approaches, it represents a widely available, easy-to-use and cost-effective entry point
for AR.

The visual aspect of the experience is important, with imaging sensors capturing traits of the
physical world allowing systems to render and overlay virtual objects onto this real world. However,
other sensory models, such as touch (haptic feedback) and sound (spatial audio) are often
important as well. This is particularly so with AR/MR in which the user may interact with digital and
real-world objects while maintaining a presence in the physical world. Gesture and posture
recognition provide hand and body tracking, and touch-sensitive feedback may be incorporated.
The emerging AR cloud will make integration of AR/VR/MR experiences richer by providing a
10
standardized spatial map of the real world.
Although the potential of VR, AR and MR is impressive, there will be many challenges and
roadblocks. Identify key target personas and explore targeted scenarios. For example, explore the
needs of, and business value for, a target user in different settings, such as at home, in a car, at
work, or with a customer. By 2022, 70% of enterprises will be experimenting with immersive
technologies for consumer and enterprise use, and 25% will have deployed them to production.
11
Three use cases show clear value:
■ Product design and visualization. This can be an internal-facing or external-facing use case.
Immersive experience technologies enable designers and customers to conceptualize, design,
explore and evaluate products virtually, without the need for physical prototypes, while taking
into consideration the users’ constraints and environment. This use delivers improved R&D
collaboration across teams, shorter product development cycles and better product designs
through simulations. This use case is prevalent in real estate and automotive development,
although it can be applied to products of all kinds.
■ Field service and operations. This is a remote and AR-/MR-based use case. Immersive
experience technologies support frontline workers completing their tasks on-site, providing
augmentation to improve efficiency. To date, this use case has been mainly for maintenance and
repair, parts visualization and visual guidance. By 2023, two out of three large field service
organizations will equip field technicians with immersive applications to improve efficiency and
customer satisfaction, up from less than 1% in 2019.
■ Training and simulation. This is an internal-facing, indoor and VR-based use case. It helps
12
employees to learn new, or improve existing, skills. It enables more flexible, efficient and self-
service training, and the use of role-play training for first responders and other frontline-facing
scenarios. This use case includes training for mission-critical tasks and advanced operational
skills. By 2022, 35% of large businesses in the training and simulation industry will evaluate and
adopt immersive solutions, up from less than 1% in 2019.
The Language Factor

A conversational platform provides a high-level design model and execution engine in which user
and machine interactions occur. As with immersive experience, other input/output (I/O) mechanisms
will be added to exploit sight, taste, smell and touch for a multisensory interaction. This is already
happening with both Google and Amazon adding screens and cameras to their smart speakers. The
use of expanded sensory channels will support capabilities, such as emotion detection through
facial expression analysis or analyzing data from accelerometers to identify abnormal movements

that may indicate a health condition. But exploitation of these other sensory channels will be
isolated and limited through 2023.
Conversational platforms have reached a tipping point: The usefulness of the systems has
exceeded the friction of using them. But they still fall short. Friction is created when users need to
know which domains the UI understands and what its capabilities are within those domains. A key
challenge that conversational platforms face is that users must communicate in a very structured
way. This is often a frustrating experience. Rather than enabling a robust two-way conversation
between the person and the computer, most conversational platforms are mainly one-directional
query or control systems that produce a very simple response. This is beginning to change as
vendors work to create a more natural conversational flow.
Over time, more conversational platforms will integrate with growing ecosystems of third-party
services that will exponentially drive the usefulness of these systems. Primary differentiators among
conversational platforms will be the robustness of their conversational models and the API and
event models used to access, invoke and orchestrate third-party services to deliver complex
outcomes.
The Multiexperience Development Platform

A multiexperience development platform (MXDP) offers front-end development tools and back-end
services that enable rapid, scalable development of seamless, targeted and ambient experiences
across devices, modalities, and touchpoints. The design time and runtime tools and services are
delivered via a unified development platform that has loosely coupled front- and back-end
architectures.
MXDPs are not “build once, run everywhere” or “omnichannel” solutions.
The core value of an MXDP lies in its ability to coalesce

software development life cycle activities across a range of
apps to address the digital user journey. The need for this ability
will only increase as the number of apps, devices and modes of
interaction increases.
The best practice for providing an optimized experience is to build connected, fit-for-purpose apps
that align the design, function and capabilities of the app to individual personas and modalities.
Unlike a one-size-fits-all application experience that attempts to support every feature that every
type of user wants from the software, fit-for-purpose apps focus in on the customer moments or
steps in the journey that users perform to attain a goal or outcome. This approach provides the
opportunity to create smaller purpose-built apps that are easier to design, develop and deploy. This
approach also allows the app to be focused on whatever touchpoint the user prefers when
performing tasks, whether it’s a web-based app running on a laptop, a mobile app on a phone or a
conversational interface.

An effortless experience should ensure that a particular persona has a consistent experience when
moving from one device to another or using multiple devices simultaneously in a more ambient
experience. Multiexperience design requires the back end to be flexible enough to support the
different capabilities and workflows of every app. Because users don’t always use the same
devices, and often switch from one device to another during their working day, the back end must
13
offer a continuous experience.
Related Research:
■ “Market Guide for Conversational Platforms”

■ “3 Immersive Experience Use Cases That Provide Attractive Market Opportunities”
■ “Survey Analysis: Insights to Kick-Start an Enterprise Multiexperience Development Strategy”
■ “Critical Capabilities for Multiexperience Development Platforms”
■ “Magic Quadrant for Multiexperience Development Platforms”
■ “Architecting and Integrating Chatbots and Conversational Platforms”
■ “Market Guide for Augmented Reality”
■ “Architecture of Conversational Platforms”
Trend No. 3: Democratization

Democratization is focused on providing people with access to technical expertise (e.g., ML,
application development) or business domain expertise (e.g., sales process, economic analysis) via
a radically simplified experience and without requiring extensive and costly training. The notion of
“citizen access” (e.g., citizen data scientists, citizen integrators) as well as the evolution of citizen
development and no-code models are examples of democratization. Development of expert
systems or virtual assistants based on AI and decision models is another important aspect of
14
democratization. These systems provide advice or take actions on behalf of people to extend their
knowledge or expertise beyond their experience or training.
It is important to note that the target for the democratization trend could be any person inside or
outside the enterprise including customers, business partners, corporate executives, salespeople,
assembly line workers, professional application developers, and IT operations professionals. There
are four key aspects to the democratization trend that are accelerating in 2020 through 2023:
■ Democratization of Application Development. AI PaaS provides access to sophisticated AI

tools to leverage custom-developed applications. These solutions provide AI-model-building
tools, APIs and associated middleware that enable the building/training, deployment and
consumption of machine learning models running on prebuilt infrastructure-as-cloud services.
These cover vision, voice, and general data classification and prediction models of any type.
■ Democratization of Data and Analytics. The tools used to build AI-powered solutions are
expanding from tools targeting data scientists (AI infrastructure, AI frameworks and AI

platforms) to tools targeting the professional developer community (AI platforms and AI
services) and the citizen data scientist. This includes tools to generate synthetic training data,
15
which helps address a substantial barrier to ML model development.
■ Democratization of Design. In addition, low-code application development platform tools
used to build AI-powered solutions are themselves being empowered with AI-driven capabilities
that assist professional developers and automate tasks related to the development of AI-
enhanced solutions. This expands on the low-code, no-code phenomenon with automation of
additional application development functions to empower the citizen developer.
■ Democratization of Knowledge. Non-IT professionals increasingly have access to powerful
tools and expert systems that empower them to exploit and apply specialized skills beyond
their own expertise and training. Dealing with the issues around “shadow AI” in this user-led
environment will be a challenge.
Democratization of Application Development

The market is rapidly shifting from one in which professional data scientists must partner with
application developers to create most AI-enhanced solutions to one in which professional
developers can operate alone using predefined models delivered as a service. This provides the
developer with an ecosystem of AI models, as well as easy-to-configure development tools tailored
to integrating AI capabilities and models into a solution.
Some AI PaaS services are complete models that a developer can simply call as a function, pass
the appropriate parameters and data, and obtain a result. Others may be pretrained to a high level
but require some additional data to customize the models. For example, a model may be pretrained
for image recognition but requires a training dataset to recognize a particular set of images. The
advantage of these partially trained models is that they require much smaller datasets of relevant
enterprise data for refined training.
Not only does the evolution of these AI platforms and suites of

AI services enable a wider range of developers to deliver AI-
enhanced solutions, but it also delivers much higher developer
productivity.
This reduces waste and inefficiency in the development life cycle of AI projects. The pretrained
models can be accessed via API calls or event triggers.
Application teams must determine which AI services to use from external providers. Next, define an
architecture for how the organization’s data science teams will develop custom domain and
company-specific AI services as part of the AI service ecosystem. This will make the cloud service
provider decision even more complex, as it requires the selection of the underlying platforms,
frameworks and infrastructure to build, train and deploy these models. We expect increasing
demand to lead to standardization of the deployment of these custom models across multiple
environments.

Low-Code, No-Code and the Citizen Developer
Low-code application development is not new, but a confluence of digital disruptions has led to an
influx of tools to meet rising demand for rapid application development platforms. There is a wide
16
range of vendors providing these solutions, spanning simple forms creation to full stack
application platforms. Low-code development offerings are part of this tool spectrum as well and
are primarily targeted at enabling citizen developers in the lines of business. By 2024, low-code
application development will be responsible for more than 65% of application development activity.
Also, by 2024, 75% of large enterprises will be using at least four low-code development tools for
both IT application development and citizen development initiatives.
Another level of democratization occurs as AI is applied to automate various application

development and testing functions. Through 2020, assisted development and testing will emerge to
simplify these functions. By 2022, we expect more mainstream use of virtual software engineers to
generate code. Google’s AutoML is one example of how augmented analytics will enable
developers to automatically generate new models without the involvement of a professional data
scientist. By 2022, at least 40% of new application development projects will have artificial
intelligence co-developers on the team.
Ultimately, highly advanced AI-powered development

environments automating both functional and nonfunctional
aspects of applications will give rise to a new age of the “citizen
application developer.”
In this new age, nonprofessionals will be able to use AI-driven tools to automatically generate new
solutions. We expect that AI-powered systems will drive a new level of flexibility. They will enable
nonprofessionals to rapidly create much more dynamic, open and complex solutions.
Augmented Analytics and the Citizen Data Scientist

Augmented analytics uses ML- and AI-assisted data preparation, insight generation, and insight
explanation to allow businesspeople to act as “citizen data scientists” exploring and analyzing data
without the assistance of professional data scientists. It democratizes analytics and AI in three key
ways:
■ Augmented data generation and preparation, which uses ML automation to augment data
creation, profiling, quality, harmonization, modeling, manipulation, enrichment, cataloging and
metadata development. This is also transforming all aspects of data management, including
automating data integration and database and data lake administration.
■ Augmented analytics as part of analytics and business intelligence (BI), which enables
business users and citizen data scientists to automatically discover, visualize and narrate
relevant findings without building models or writing algorithms. These findings may include

correlations, exceptions, clusters, segments, outliers and predictions. Users explore data via
autogenerated visualizations and conversational interfaces.
■ Augmented analytics uses ML to automate aspects of data science and AI modeling, such
as feature engineering, automated machine learning (autoML) for model selection, model
operationalization, model explanation and, ultimately, model tuning and management. This
reduces the need for specialized skills to generate, operationalize and manage models.
Organizations can use citizen data scientists or semitrained business experts to fill the data science
and ML talent gap. By 2021, automation of data science tasks will enable citizen data scientists to
produce a higher volume of advanced analysis than specialized data scientists. Gartner predicts
that, by 2021, augmented analytics will be a dominant driver of new purchases of analytics and BI
— as well as data science and machine learning — platforms, and of embedded analytics. We also
expect that, by 2025, a scarcity of data scientists will no longer hinder the adoption of data science
and machine learning in organizations. Organizations increasingly augment and outsource data
science work.
Augmented analytics will also be a key feature of analytics embedded in autonomous things that
interact with users — especially autonomous assistants using conversational interfaces. This
emerging way of working enables businesspeople to generate queries, explore data, and receive
and act on insights in natural language (voice or text) via mobile devices and personal assistants.
However, this is only the initial use of augmented analytics in autonomous things. Augmented
analytics enables the embedding of an augmented data science function in any type of autonomous
thing. When the autonomous thing requires analytics to operate, it can tap into the embedded
augmented analytics function to respond.
In a typical analytics scenario, people often default to exploring their own biased hypotheses,
missing key findings and drawing their own incorrect or incomplete conclusions. This may adversely
affect decisions and outcomes. Augmented analytics enables the exploration of more hypotheses
and the identification of hidden patterns. It also removes personal bias. However, AI algorithms are
not inherently unbiased, and care must be taken not to inadvertently introduce new biases through
17
limited training sets.
Dealing With Shadow AI

Shadow AI refers to a natural consequence of democratization where individuals without formal
training exploit easy-to-use tools to develop their own AI-powered solutions and provide peer-to-
peer support to others in similar efforts. Like “shadow IT” where business users bring their own
consumer technologies to work or create their own spreadsheet or analytical models with BI tools,
“shadow AI” has both a positive and a negative side. Democratization opens up new opportunities
for the citizen data scientist, integrator and developer. Business users can dynamically create
powerful AI-powered models and analytics with increasingly easy-to-use tools. While this can be a
productivity bonanza and enable the business to more-rapidly adapt and drive new business
opportunities, there is also a challenge with unleashing these powerful tools on an untrained
audience.

Shadow AI takes “bring your own” to a more granular level by allowing “bring your own data” and
“bring your own algorithm” outside the ownership, control or stewardship of IT. Shadow AI is not an
inherently bad thing as long as practices and training are in place. By 2022, 30% of organizations
using AI for decision making will contend with shadow AI as the biggest risk to effective and ethical
decisions.
Related Research:
■ “Predicts 2019: The Democratization of AI”

■ “Low-Code Development Technologies Evaluation Guide”
■ “Four Real-World Case Studies: Implement Augmented DSML to Enable Expert and Citizen
Data Scientists”
■ “Innovation Insight for AI-Augmented Development”
■ “Augmented Analytics Feature Definition Framework”
■ “Top 10 Data and Analytics Technology Trends That Will Change Your Business”
■ “How to Enable Self-Service Analytics”
■ “Predicts 2019: Democratization of IT Requires Different Strategies for Integration”
Trend No. 4: Human Augmentation

Human augmentation refers to the enhancement of human capabilities and capacity through the
use of technology and science. Humans have always used technology and science in this fashion.
Even before the introduction of the computer, technologies such as the typewriter, copy machine
and printing press augmented the human ability to create, copy and publish text. Glasses, hearing
aids and false teeth are all historical examples of human augmentation.
The computer era has added new dimensions to human augmentation. Word processing, desktop
publishing, webpages, blogs and social media greatly extend our ability to create and publish text.
With the rise of new technologies such as IoT, AI, smart speakers and VR emerging from computer
science, and technologies such as CRISPR18 emerging from biological science, entirely new
opportunities for human augmentation are emerging.
Human augmentation explores how technology can be used to deliver cognitive and physical
improvements as an integral part of the human experience. Instead of computers and applications
being something outside the normal human experience, they become a natural — and sometimes
necessary — part of the day-to-day human experience. Moreover, human augmentation also
includes bioengineering factors that go beyond exploitation of computers and applications. We are
already on this path to a certain extent. For many people, smartphones are an essential tool and
constant companion. Social networks and electronic connections like email have become a primary
link between people. Pharmaceuticals have augmented humans since well before the advent of
computers. Human augmentation is a prime example of combinatorial innovation which brings
together many trends including:

■ Hyperautomation and the development of expert systems to democratize access to skills
beyond current experience and training.
■ Empowered edge devices and autonomous things, which exist in spaces around humans and
augment their capabilities.
Cognitive and Physical Augmentation

Human augmentation impacts how we move, perceive, and interact in both physical and digital
spaces, as well as how we process, analyze and store information. Augmentation can be broadly
categorized into physical and cognitive categories, although the boundaries between them will blur
over time.
Physical augmentation enhances humans by changing their inherent physical capabilities by

implanting or hosting a technology element on their bodies. Automotive, mining, oil and gas, and
other industries are using wearables to improve worker safety. Wearables are also driving workplace
productivity in industries like retail, travel and healthcare. Physical augmentation also includes using
biology or other means to alter the human body. In some cases, physical augmentation replaces a
human capability that an individual has lost (e.g., a prosthetic leg); but, in some cases, those
replacement capabilities may surpass natural human abilities. Physical augmentation can be
considered along a number of dimensions:
■ Sensory Augmentation — Hearing, vision and other augmentation devices or implants to

improve perception. Virtual, augmented and mixed reality are a current example of sensory
augmentation. In the emerging technology space, various companies are experimenting with
19
smart contact lenses to detect glucose levels in tears and intraocular pressure. Researchers
20
are also experimenting with developing an “electronic nose” that mimics a human nose.
■ Appendage and Biological Function Augmentation — Use of exoskeletons and prosthetics
to replace or enhance these capabilities is an expanded area of human augmentation. Surgical
21
augmentation of the eyes has been popular with professional golfers. Cochlear implants can
replace nonfunctioning auditory nerves, and a similar technology has been used to replicate
eyes. The cosmetics industry leads in enhancing nails, hair, eyes and the shape of body parts
using passive implants. Arguably, the pharmaceutical industry has been augmenting human
biological functions for years. Nootropics refers to the use of natural or synthetic substances
that may improve mental skills, although the use of such substances outside of treatment for a
specific medical condition is highly controversial.
■ Brain Augmentation — Implants such as a vagus nerve stimulator to treat seizures currently
22 23
exist. Brain implants are being explored for a variety of uses including memory storage and
24
brain implants to decode neural patterns and synthesize speech. Neuralink is trying to develop
a brain implant that would connect the human brain to computer networks.25
■ Genetic Augmentation — Somatic gene and cell therapies are used today and seen as morally
acceptable. For example, gene therapy to treat children with severe combined immune
deficiency is an accepted treatment. In the future, easy access to and the low cost of CRISPR
26
technologies may enable broad genetic engineering, although the ethical issues are significant.

Cognition is the process by which humans acquire knowledge through sensory input, life
experiences, learning, and thinking about that input, experience and education. Cognitive skills are
used to understand, process, remember, and apply information to make decisions and take actions.
Cognitive augmentation enhances a human’s ability to think and make better decisions. Cognitive
augmentation can occur through accessing information and exploiting applications on traditional
computer systems and the emerging multiexperience interface in smart spaces. This includes
augmented intelligence scenarios, where humans and artificial intelligence work together to
enhance cognitive performance, including decision making and learning. In addition, physical
augmentation that enhances human senses or brain capacity or capability feed into new models for
cognitive augmentation. This includes the use of smart drugs and brain implants to store
23
memories.
Human augmentation offers the opportunity to achieve digital

transformation through human transformation.
Wearables are an example of physical augmentations today. As the maturity and adoption of
wearables increase, consumers and employees will begin looking at other physical augmentations
to enhance their personal lives (i.e., health and fitness) or to do their jobs more efficiently (i.e.,
exoskeletons and implants). Over the next 10 years, increasing levels of physical and cognitive
human augmentation will become prevalent as individuals seek personal enhancements. This in turn
will create a new “consumerization” effect where employees seek to exploit their personal
enhancements — and even extend them — to improve their office environment. Through 2023, 30%
of IT organizations will extend BYOD policies with “bring your own enhancement” to address
augmented humans in the workforce.
Cultural and Ethical Aspects of Human Augmentation

Human augmentation will be a primary means by which individuals interact with each other and with
the smart spaces that surround them. Business leaders and IT leaders should plan for how their
organizations will adopt, exploit and adapt to the coming changes. As consumers and employees
integrate more of their lives into one intelligence-amplifying human augmentation, organizations will
27
have to address issues of data transparency, privacy and autonomy.
When choosing human augmentation technologies and methodologies, enterprises must examine
five major areas:
■ Security. Human augmentation technologies must achieve and maintain a known and
acceptable state of security-related risk. This risk is across an attack surface that’s no longer
tied to a specific device or physical location, but may travel with the human subject.
■ Privacy. Human augmentation provides the ability to access intimate knowledge and data
about the human it’s enhancing. That data must be protected.
■ Compliance. Government and regulatory agencies are issuing regulations and providing
compliance requirements on a frequent basis, which makes compliance extremely complex for

global enterprises, especially because agencies are still trying to grasp the implications of
human augmentation technologies.
■ Health impact. Human augmentation has the potential for long-term mental and physical
implications that may not be immediately understood.
■ Ethics. Implementing human augmentation technologies and processes poses serious ethical
issues. These include ethical considerations and assessments for determining specific
vulnerabilities, risks and moral issues. For example, does the digital divide widen as affluent
individuals can augment themselves and their children while less affluent people cannot?
Answers to these societal issues will become increasingly important.
Enterprises of all types and sizes are considering human augmentation to achieve various business
outcomes through multiple business use cases with different time horizons. Consequently, they
must consider the lessons, recommendations and principles of human experimentation as they
intentionally begin to exploit human augmentation capabilities for human transformation.
Enterprises should strike a balance between two classic ethical principles — precautionary and
proactionary — and adopt what Gartner calls the “cautionary principle”:
■ The precautionary principle states that “if an action or policy might cause severe or irreversible
harm to the public or to the environment, in the absence of a scientific consensus that harm
28
would not ensue, the burden of proof falls on those who would take the action.”
■ The proactionary principle was formulated by Max More and is a key tenet of the
“transhumanism” movement. It presents several imperatives when imposing restrictive
measures: “Assess risks and opportunities according to available science, not popular
perception. Account for both the costs of the restrictions themselves, and those of
opportunities foregone. Favor measures that are proportionate to the probability and magnitude
of impacts, and that have a high expectation value. Protect people’s freedom to experiment,
29
innovate, and progress.”
■ The cautionary principle balances between precautionary and proactionary. It recommends
that organizations move forward with innovation, but only in a way that does not risk the
individual, company or environment as a whole.
Related Research:
■ “Maverick* Research: Architecting Humans for Digital Transformation”

■ “Emerging Technology Analysis: Smart Wearables”
■ “Technology Investments for Frontline Workers Will Drive Real Business Benefits”
■ “Hype Cycle for Emerging Technologies, 2019”
■ “Market Insight: Increase User Engagement for Voice-Enabled Virtual Assistants Through a
More Targeted User Experience”

Trend No. 5: Transparency and Traceability
Digital ethics and privacy are growing concerns for individuals, organizations and governments.
Consumers are increasingly aware their personal information is valuable and are demanding control.
Organizations recognize the increasing risk of securing and managing personal data, and
governments are implementing strict legislation to ensure they do.
Artificial Intelligence and the use of ML models to make autonomous decisions raises a new level of
concern, with digital ethics driving the need for explainable AI and the assurance that the AI system
is operating in an ethical and fair manner. Transparency and traceability are critical elements to
support these digital ethics and privacy needs.
Transparency and traceability are not a single product or a single action. It refers to a range of
attitudes, actions, and supporting technologies and practices designed to address regulatory
requirements, enshrine an ethical approach to use of AI and other advanced technologies, and
repair the growing lack of trust in companies.
Transparency and traceability require a focus on six key elements of trust:
■ Ethics — Does the organization have strong moral principles on the use of personal data,
algorithms and the design of systems that go beyond regulations and are transparent to all
interested parties?
■ Integrity — Does the organization have a proven track record of designing systems that reduce
or eliminate bias and inappropriate use of personal data?
■ Openness — Are the ethical principles and privacy commitments clear and easily accessible —
and do changes to such policies bring the appropriate constituencies into the decision-making
process?
■ Accountability — Are there mechanisms for testing, assurance and auditability such that
privacy or ethical concerns can be identified and addressed? This applies not only to adherence
to regulatory requirements, but also to new ethical or privacy concerns that arise from future
technologies.
■ Competence — Has the organization implemented design principles, processes, testing and
training so that concerned constituencies can feel comfortable that the organization can
execute on its promises?
■ Consistency — Are policies and processes handled consistently?
Explainable and Ethical AI and Algorithms

Algorithmic decisions drive everything organizations and consumers do: recruiting, buying products
and services, what content they see on the web, being accepted or rejected for loans, even going to
30
jail. Biased or incorrect algorithms promote bad business decisions and risk serious backlash from
employees, investors and customers. They could potentially even cause serious harm and result in
corporations facing criminal penalties. Legal risks include gender and racial bias and other forms of

discriminatory activities. Lack of explainability or an inability to prove lack of bias means business
activities can’t be justified in public or in court.
Opaque algorithms (deep neural networks [DNNs], for example) incorporate many implicit, highly
variable interactions into their predictions that can be difficult to interpret. Bias in AI raises concerns
of accountability and fairness. Consequently, the AI community and enterprise leaders are
concerned with detecting and explaining the consequences of bias that might endanger society and
the business. For example, bias in AI causes polarization of political opinions, persistence of
discredited beliefs and false associations between business moments.
Explainable AI is the set of capabilities that describes a model, highlights its strengths and
weaknesses, predicts its likely behavior, and identifies any potential biases. It can articulate the
decisions of a descriptive, predictive or prescriptive model to enable accuracy, fairness,
accountability, stability and transparency in algorithmic decision making. Explainable AI provides the
technical foundation to support AI governance. AI governance is the process of assigning and
assuring organizational accountability, decision rights, risks, policies and investment decisions for
applying AI, predictive models and algorithms.
Augmented analytics solutions with explainable AI features are not only showing data scientists the
input and output of a model, but are also explaining why the system selected particular models and
what are the techniques applied by augmented data science and ML. Without acceptable
explanation, autogenerated insights and models, or black boxes combined with AI bias, can cause
concerns about regulation, reputation, accountability and fairness, and lead to distrust in AI
solutions.
Transparency and traceability are something that enterprises need to embrace themselves, but they
will become increasingly important aspects of how organizations evaluate the packaged
applications and services they purchase using AI. By 2025, 30% of government and large-
enterprise contracts for purchase of digital products and services using AI will require the use of
explainable and ethical AI.
Data Privacy, Ownership and Control

People are increasingly concerned about how their personal information is being used by
organizations in both the public and private sector; and the backlash will only increase for
organizations that are not proactively addressing these concerns.
While the private sector is increasingly bound by privacy legislation, law enforcement and security
services have far fewer controls. Police services use facial recognition to identify people of interest.
They use automatic number plate recognition (ANPR) to track vehicles of interest. They also use
31
data from fitness trackers to establish people’s location and heart rate at the time of a crime.
32
They’ve even used Face ID to unlock a suspect’s phone. With billions of endpoints collecting
information, law enforcement can identify who you are, where you are, what you’re doing and even
what you’re thinking.

Any discussion on privacy must be grounded in the broader
topic of digital ethics and the trust of your customers,
constituents and employees. While privacy and security are
foundational components in building trust, trust is actually
about more than just these components.
As defined by Oxford Dictionaries, trust is a firm belief in the reliability, truth or ability of someone or
something. Trust is often seen as the acceptance of the truth of a statement without evidence or
investigation; but, in reality, trust is typically earned from demonstrable actions over time and not
provided blindly.
Ultimately an organization’s position on privacy must be driven by its broader position and actions
on ethics and trust. Shifting from privacy to ethics moves the conversation beyond “are we
compliant?” toward “are we doing the right thing?” and “are we showing that we are trying to do the
right thing?” The move from compliance-driven organizations to ethics-driven organizations can be
33
described as the hierarchy of intent.
People are justifiably concerned about the use of their personal data and are starting to fight back.
Misguided personalization attempts, media coverage and lawsuits have made one thing clear to
customers: Their data is valuable. And so, they want to take back control. Customers may opt out
of services, pay in cash or bitcoin, use VPNs to mask their location, provide false information or
simply fade out of the relationship.
Right to be forgotten (RTBF) legislation exists in many jurisdictions including Europe, South Africa,
South Korea and China. The right to data portability empowers customers to more easily take their
personal data — and business — elsewhere. The highly valuable personal information that
organizations have spent a decade to effectively leverage is disappearing. Failure to incorporate
privacy into a personalization strategy can bring unwanted results, such as customer churn, lack of
loyalty and distrust, as well as brand reputational damage. In some cases, regulatory intervention
could occur if customers feel their privacy is being threatened.
Companies that misuse personal data will lose the trust of their customers. Trustworthiness is a key
factor in driving revenue and profitability. Building customer trust in an organization is difficult, but
losing it is easy. However, organizations that gain and maintain the trust of their customers will
thrive. We expect that companies that are digitally trustworthy will generate more online profit than
those that aren’t. However, not all of the transparency and trust elements can be easily automated.
Greater human effort will be required to provide accountability, and points of contact for questions
and to redress when things go wrong.

Best Practices and Regulations Are Emerging
The EU’s General Data Protection Regulation (GDPR) redefines the ground rules for privacy and has
had a global impact. It allows for fines of up to 4% of annual global revenue or €20 million,
whichever is highest. We expect that, before year-end 2021, more than a billion euros in sanctions
for GDPR noncompliance will have been issued. Many other nations are in the process of
developing or implementing privacy legislation, and this evolving patchwork of privacy laws around
the world will continue to challenge organizations in the way they interact with customers, citizens
and employees.
Legislation is also driving data residency issues in jurisdictions including China, Russia, Germany
and South Korea. Organizations must assess the data residency requirements of the countries in
which they operate to determine a data residency strategy. Local data centers are an option, but
often an expensive one; and, in many cases, there are legal and logical controls available under
which safe cross-border transfers of personal data can be possible. Cloud service providers are
locating data centers in countries where data residency is either driven by legislation or by customer
preferences.
Explainable and ethical AI is becoming a political and regulatory issue. The U.K. government Centre
for Data Ethics and Innovation has started a review of algorithmic bias in financial services industry.
U.S. politicians are proposing regulations as well.
Implementing the principles of “privacy by design” positions your products and services as more
privacy-friendly than those of competitors. This creates a value proposition based on trust. A 2017
survey indicates 87% of consumers say they will take their business elsewhere if they don’t trust
34
that a company is handling their data responsibly. Other standards groups such as the IEEE
Standards Association have been building a variety of standards for ethical considerations in AI and
35
autonomous systems as well as standards for personal data and AI agents. Enterprises should
track the advancement of these principles and adopt them as needed to demonstrate ethics,
consistency and competency.
Related Research:
■ “Top 10 Data and Analytics Technology Trends That Will Change Your Business”
■ “Build Trust With Business Users by Moving Toward Explainable AI”
■ “Hype Cycle for Artificial Intelligence, 2019”
■ “Build AI-Specific Governance on Three Cornerstones: Trust, Transparency and Diversity”
■ “Modern Privacy Regulations Could Sever or Strengthen Your Ties With Customers”
■ “The CIO’s Guide to Digital Ethics: Leading Your Enterprise in a Digital Society”
■ “Use These Privacy Deliverables in Every IT Development Project”
■ “Build for Privacy”
■ “Hype Cycle for Privacy, 2019”

Trend No. 6: The Empowered Edge
Edge computing describes a computing topology in which information processing and content
collection and delivery are placed closer to the sources, repositories and consumers of this
information. Edge computing draws from the concepts of distributed processing. It tries to keep the
traffic and processing local to reduce latency, exploit the capabilities of the edge and enable greater
autonomy at the edge.
Much of the current focus on edge computing comes from the need for IoT systems to deliver
disconnected or distributed capabilities into the embedded IoT world for specific industries such as
manufacturing or retail. However, edge computing will come to be a dominant factor across virtually
all industries and use cases as the edge is empowered with increasingly more sophisticated and
specialized compute resources and more data storage. Increasingly complex edge devices
including robots, drones, autonomous vehicles and operational systems are accelerating this shift in
focus.
The evolution of edge-oriented IoT architectures, with intelligence migrating toward endpoints,
gateways and similar devices, is underway. However, today’s edge architectures are still somewhat
hierarchic, with information flowing through well-defined layers of endpoints to the near edge,
sometimes the far edge and, eventually, centralized cloud and enterprise systems.
Over the long term, this neat set of layers will dissolve to create a more unstructured architecture
consisting of a wide range of “things” and services connected in a dynamic flexible mesh linked by
a set of distributed cloud services. In this scenario, a smart “thing,” such as a drone, might
communicate with an enterprise IoT platform, a government drone tracking service, local sensors
and city-level local cloud services, and then conduct peer-to-peer exchanges with nearby drones
for navigational purposes.
The edge, near edge and far edge connect to centralized data centers and cloud services. Edge
computing solves many pressing issues, such as high bandwidth costs and unacceptable latency.
The edge computing topology will enable the specifics of digital business and IT solutions uniquely
well in the near future.
The evolution of distributed cloud will increasingly provide a

common or complementary set of services that can be centrally
managed but delivered to the edge environment for execution.
Mesh architectures will enable more flexible, intelligent and responsive, and peer-to-peer IoT
systems — although often at the cost of additional complexity. A mesh architecture is also a
consequence of many distributed networked ecosystems. Fundamental shifts to digital business
and products will have to exploit intelligent mesh architectures for competitive advantage. The
evolution from centralized to edge to mesh will also have a major impact on product development
and evolving teams’ skills on both cloud and edge designs. Mesh standards are immature, although
groups such as the IEEE and the OpenFog Consortium are working in the mesh architecture area.

As endpoints grow in number and become more sophisticated, AI-driven, and able to run operating
systems such as Linux, mesh architectures will become more popular. New mesh architectures will
introduce significant complexity into IoT systems and will likely make tasks such as designing,
testing and support more challenging. In addition, the mesh often implies more peer-to-peer
activities, which will involve partners and ecosystems stretching beyond individual products.
Through 2028, we expect a steady increase in the embedding of sensor, storage, compute and
advanced AI capabilities in edge devices. However, the edge is a heterogeneous concept. It ranges
from simple sensors and embedded edge devices to familiar edge computing devices, such as
mobile phones, and highly sophisticated edge devices, such as autonomous vehicles. Different
types of edge devices used in different scenarios can have very different life spans, ranging from
one year to 40 years. These factors combined with the rapid push by vendors to drive more
functionality into edge devices create a complex and ongoing management and integration
challenge.
Intelligence will move to the edge across a spectrum of endpoint devices including:
■ Simple embedded edge devices (e.g., appliances, industrial devices)

■ Edge input/output devices (e.g., speakers, screens)
■ Edge computing devices (e.g., smartphones, PCs)
■ Complex embedded edge devices (e.g., automobiles, power generators)
These edge systems will connect with hyperscale back-end services directly or through
intermediary edge servers or gateways (see Note 3).
Data, Analytics and AI at the Edge

By 2022, as a result of digital business projects, 75% of enterprise-generated data will be created
and processed outside the traditional, centralized data center or cloud — an increase from the less
than 10% generated today. This move toward distributed data forces organizations to strike a
different balance between collecting data centrally for processing and connecting to the data where
it is generated to perform local processing. Modern use cases demand that data management
capabilities move toward the edge — bringing the processing to the data, rather than always
collecting the data for centralized processing. But the distributed “connect” paradigm brings a lot of
challenges with it. For example:
■ If data persists at the edge, then in what form will it be stored?

■ How will governance controls be enforced there?
■ How will it be integrated with other data?
Data and analytics leaders need to modernize how data assets are described, organized, integrated,
shared and governed to address these challenges.
Although not all aspects of data management will need to happen at the edge, modern use cases
increasingly will push requirements in that direction (see “Technology Insight: Edge Computing in

Support of the Internet of Things”). The highly distributed nature of modern digital business
applications, including architectures for IoT solutions, promises to challenge organizations’ abilities
to manage and process data at a level of scale and complexity for which most are unprepared. This
36
shift to edge computing will have several impacts:
■ Data and analytics use cases and solutions require support for new distributed data architecture
that is beyond data and analytics leaders’ current data management capabilities.
Include distributed data stores and processing in your data management to ensure you can
offer support where data resides.
■ Distributed data requires distributed data management capabilities, compelling data and
analytics leaders to rebalance their ability to bring processing to data at the edge.
Extend data persistence capabilities with cloud-based data stores, distributed parallel
processing platforms and embedded database technology.
■ Edge computing and other distributed environments will challenge data management
technology provider capabilities, resulting in data and analytics leaders more closely scrutinizing
how they navigate related technology markets.
Assess incumbent and prospective vendors based on ability to process and govern distributed
data.
Communicating to the Edge — The Role of 5G

Connecting edge devices with one another and with back-end services is a fundamental aspect of
IoT and an enabler of smart spaces. 5G is the next-generation cellular standard after 4G Long Term
Evolution (LTE; LTE Advanced [LTE-A] and LTE Advanced Pro [LTE-A Pro]). Several global standards
bodies have defined it — International Telecommunication Union (ITU), 3rd Generation Partnership
Project (3GPP) and ETSI. Successive iterations of the 5G standard also will incorporate support for
NarrowBand Internet of Things (NB-IoT) aimed at devices with low-power and low-throughput
requirements. New system architectures include core network slicing as well as edge computing.
5G addresses three key technology communication aspects, each of which supports distinct new
services, and possibly new business models (such as latency as a service):
■ Enhanced mobile broadband (eMBB), which most providers will probably implement first.
■ Ultrareliable and low-latency communications (URLLC), which addresses many existing
industrial, medical, drone and transportation requirements where reliability and latency
requirements surpass bandwidth needs.
■ Massive machine-type communications (mMTC), which addresses the scale requirements of IoT
edge computing.
Use of higher cellular frequencies and massive capacity will require very dense deployments with
higher frequency reuse. As a result, we expect that most public 5G deployments will initially focus
on islands of deployment, without continuous national coverage. We expect that, by 2020, 4% of

network-based mobile communications service providers globally will launch the 5G network
commercially. Many CSPs are uncertain about the nature of the use cases and business models
that may drive 5G. We expect that, through 2022, organizations will use 5G mainly to support IoT
communications, high-definition video and fixed wireless access. The release of unlicensed radio
spectrum (Citizens Broadband Radio Service [CBRS] in the U.S., and similar initiatives in the U.K.
and Germany) will facilitate the deployment of private 5G (and LTE) networks. This will enable
enterprises to exploit the advantages of 5G technology without waiting for public networks to build
out coverage.
Identify use cases that definitely require the high-end performance, low latency or higher densities
of 5G for edge computing needs. Map the organization’s planned exploitation of such use cases
against the expected rollout by providers through 2023. Evaluate the available alternatives that may
prove adequate and more cost-effective than 5G for particular IoT use cases. Examples include
low-power wide-area (LPWA), such as 4G LTE-based NB-IoT or LTE Cat M1, LoRa, Sigfox and
Wireless Smart Ubiquitous Networks (Wi-SUN).
Digital Twin of Things at the Edge

A digital twin is a digital representation of a real-world entity or system. The implementation of a
digital twin is an encapsulated software object or model that mirrors a unique physical object (see
Note 4). Data from multiple digital twins can be aggregated for a composite view across a number
of real-world entities such as a power plant or a city.
Well-designed digital twins of assets could significantly improve enterprise decision making. They
are linked to their real-world counterparts at the edge and are used to understand the state of the
thing or system, respond to changes, improve operations and add value.
Organizations will implement digital twins simply at first and evolve them over time, improving their
ability to collect and visualize the right data, apply the right analytics and rules, and respond
effectively to business objectives. Digital-twin models will proliferate, with suppliers increasingly
providing customers with these models as an integral part of their offerings.
Related Research:
■ “The Future Shape of Edge Computing: Five Imperatives”

■ “Ask These Four Questions About Enterprise 5G”
■ “Digital Business Will Push Infrastructures to the Edge”
■ “Five Approaches for Integrating IoT Digital Twins”
■ “Why and How to Design Digital Twins”
■ “The Technical Professional’s Guide to Edge Infrastructure”
■ “How Edge Computing Redefines Infrastructure”

Trend No. 7: Distributed Cloud
A distributed cloud refers to the distribution of public cloud services to different locations outside
the cloud providers’ data centers, while the originating public cloud provider assumes responsibility
for the operation, governance, maintenance and updates. This represents a significant shift from the
centralized model of most public cloud services and will lead to a new era in cloud computing.
Cloud computing is a style of computing in which elastically scalable IT-enabled capabilities are
delivered as a service using internet technologies. Cloud computing has long been viewed as
synonymous with a “centralized” service running in the provider’s data center; although, private and
hybrid cloud options emerged to complement this public cloud model. Private cloud refers to the
creation of cloud-style services dedicated to individual companies often running in their own data
centers. Hybrid cloud refers to the integration of private and public cloud services to support
parallel, integrated or complementary tasks. The aim of the hybrid cloud was to blend external
services from a provider and internal services running on-premises in an optimized, efficient and
cost-effective manner.
Implementing a private cloud is hard. Most private cloud projects do not deliver the cloud outcomes
37
and benefits organizations seek. Also, most of the conversations Gartner has with clients about
hybrid cloud are actually not about true hybrid cloud scenarios. Instead, they are about hybrid IT
scenarios in which noncloud technologies are used in combination with public cloud services in a
spectrum of cloud-like models. Hybrid IT and true hybrid cloud options can be valid approaches,
and we recommend them for certain critical use cases. However, most hybrid cloud styles break
many of the cloud computing value propositions, including:
■ Shifting the responsibility and work of running hardware and software infrastructure to cloud
providers
■ Exploiting the economics of cloud elasticity (scaling up and down) from a large pool of shared
resources
■ Benefiting from the pace of innovation in sync with the public cloud providers
■ Using the cost economics of global hyperscale services
■ Using the skills of large cloud providers in securing and operating world-class services
Distributed Cloud Delivers on the Hybrid Cloud Promise

The location of the cloud services is a critical component of the distributed cloud computing model.
Historically, location has not been relevant to cloud definitions, although issues related to it are
important in many situations. With the arrival of the distributed cloud, location formally enters the
definition of a style of cloud services. Location may be important for a variety of reasons, including
data sovereignty and latency-sensitive use cases. In these scenarios, the distributed cloud service
provides organizations with the capabilities of a public cloud service delivered in a location that
meets their requirements.

In hyperscale public cloud implementations, the public cloud is the “center of the universe.”
However, cloud services have been distributed worldwide in the public cloud almost since its
inception. The providers now have different regions around the world, all centrally controlled,
managed and provided by the public cloud provider. The distributed cloud extends this model
outside cloud-provider-owned data centers. In the distributed cloud, the originating public cloud
provider is responsible for all aspects of cloud service architecture, delivery, operations, governance
and updates. This restores cloud value propositions that are broken when customers are
responsible for a part of the delivery, as is usually the case in hybrid cloud scenarios. The cloud
provider does not need to own the hardware on which the distributed cloud service is installed. But
in a full implementation of the distributed cloud model, the cloud provider must take full
responsibility for how that hardware is managed and maintained.
We expect distributed cloud computing will happen in three phases:
■ Phase 1: A like-for-like hybrid mode in which the cloud provider delivers services in a
distributed fashion that mirror a subset of services in its centralized cloud for delivery in the
enterprise.
■ Phase 2: An extension of the like-for-like model in which the cloud provider teams with third
parties to deliver a subset of its centralized cloud services to target communities through the
third-party provider. An example is the delivery of services through a telecommunications
provider to support data sovereignty requirements in smaller countries where the provider does
not have data centers.
■ Phase 3: Communities of organizations share distributed cloud substations. We use the term
“substations” to evoke the image of subsidiary stations (like branch post offices) where people
gather to use services. Cloud customers can gather at a given distributed cloud substation to
consume cloud services for common or varied reasons if it is open for community or public use.
This improves the economics associated with paying for the installation and operation of a
distributed cloud substation. As other companies use the substation, they can share the cost of
the installation. We expect that third parties such as telecommunications service providers will
explore the creation of substations in locations where the public cloud provider does not have a
presence. If the substation is not open for use by others outside the organization that paid for
its installation, then the substation represents a private cloud instance in a hybrid relationship
with the public cloud.
The distributed cloud supports continuously connected and intermittently connected operation of
like-for-like cloud services from the public cloud “distributed” to specific and varied locations. This
enables low-latency service execution where the cloud services are closer to the point of need in
remote data centers or all the way to the edge device itself. This can deliver major improvements in
performance and reduce the risk of global network-related outages, as well as support occasionally
connected scenarios. By 2024, most cloud service platforms will provide at least some services that
execute at the point of need.
Getting to Distributed Cloud

The distributed cloud is in the early stages of development. Many providers aim to offer most of
their public services in a distributed manner in the long term. But for now, they provide only a

subset — and often a small subset — of their services in a distributed way. Some of the providers’
approaches do not support the full delivery, operation and update elements of a full distributed
cloud. Providers are extending services to third-party data centers and out to the edge with
offerings such as Microsoft Azure Stack, Oracle Cloud at Customer, Google’s Anthos, IBM Red Hat
and the previously announced AWS Outposts. Enterprises should evaluate the potential benefits
and challenges of three approaches they are likely to see:
■ Software: The customer buys and owns a hardware platform and software layer with a subset
of the providers’ cloud services. The provider does not take responsibility for the ongoing
operations, maintenance or update of the software or the underlying hardware platform. Users
are responsible, doing it themselves or using a managed service provider. Although a software
approach provides a like-for-like model between the public service and the on-premises
implementation, the other challenges with the hybrid cloud remain. Some customers consider it
an advantage that they are in control of service updates.
■ Portability Layer: The provider delivers a portability layer typically built on Kubernetes as the
foundation for services across a distributed environment. In some cases, the portability layer is
simply using containers to support execution of a containerized application. In other cases, the
provider delivers some of its cloud services as containerized services that can run in the
distributed environment. The portability approach ignores the ownership and management of
the underlying hardware platform, which remains the responsibility of the customer.
■ Distributed Service: The provider delivers a like-for-like version of some of its cloud services in
a hardware/software combination, and the provider commits to managing and updating the
service. This reduces the burden on the consumer who can view the service as a “black box.”
However, some customers will be uncomfortable giving up all control of the underlying
hardware and software update cycles. Nevertheless, we expect this approach will dominate
over time.
The Edge Effect

The fundamental notion of the distributed cloud is that the public cloud provider is responsible for
the design, architecture, delivery, operation, maintenance, updates and ownership, including the
underlying hardware. However, as solutions move closer to the edge, it is often not desirable or
feasible for the provider to own the entire stack of technology. As these services are distributed onto
operational systems (for example, a power plant), the consuming organization will not give up
ownership and management of the physical plant to an outsider provider. However, the consuming
organization may be interested in a service that the provider delivers, manages and updates on
such equipment. The same is true for mobile devices, smartphones and other client equipment. As
a result, we expect a spectrum of delivery models with the provider accepting varying levels of
ownership and responsibility.
Another edge factor that will influence the distribution of public cloud services will be the
capabilities of the edge, near-edge and far-edge platforms that may not need or cannot run a like-
for-like service that mirrors what is found in the centralized cloud. Complementary services tailored
to the target environment, such as a low-function IoT device, will be part of the distributed cloud

spectrum (e.g., AWS IoT Greengrass). However, at a minimum, the cloud provider must design,
architect, distribute, manage and update these services if they are to be viewed as part of the
distributed cloud spectrum.
Related Research:
■ “The Edge Completes the Cloud: A Gartner Trend Insight Report”

■ “Hype Cycle for Cloud Computing, 2019”
■ “Define and Understand New Cloud Terms to Succeed in the New Cloud Era”
■ “Prepare for AWS Outposts to Disrupt Your Hybrid Cloud Strategy”
■ “Rethink Your Internal Private Cloud”
■ “When Private Cloud Infrastructure Isn’t Cloud, and Why That’s Okay”
■ “Cloud Computing Primer for 2019”
Trend No. 8: Autonomous Things

Autonomous things are physical devices that use AI to automate functions previously performed by
humans. The most recognizable forms of autonomous things are robots, drones, autonomous
vehicles/ships and appliances. AI-powered IoT elements, such as industrial equipment and
consumer appliances, are also a type of autonomous thing. Each physical device has a focus for its
operation as it relates to humans. Their automation goes beyond the automation provided by rigid
programming models, and they exploit AI to deliver advanced behaviors that interact more naturally
with their surroundings and with people. Autonomous things operate across many environments
(land, sea and air) with varying levels of control. Autonomous things have been deployed
successfully in highly controlled environments such as mines. As the technology capability
improves, regulation permits and social acceptance grows, autonomous things will increasingly be
deployed in uncontrolled public spaces.
Common Technology Capabilities

Autonomous things are developing very rapidly, partly because they share some common
technology capabilities. Once the challenges to developing a capability have been overcome for one
type of autonomous thing, the innovation can be applied to other types of autonomous things. The
following common capabilities and technologies were inspired by the 2020 NASA Technology
38
Taxonomy:
■ Perception: The ability to understand the physical space in which the machine is operating.
This includes the need to understand the surfaces in the space, recognize objects and their
trajectories, and interpret dynamic events in the environment.
■ Interaction: The ability to interact with humans and other things in the physical world using a
variety of channels (such as screens and speakers) and sensory outputs (such as light, sound
and haptics).

■ Mobility: The ability to safely navigate and physically move from one point to another in the
space through some form of propulsion (such as walking, cruising/diving, flying and driving).
■ Manipulation: The ability to manipulate objects in the space (such as lifting, moving, placing
and adjusting) and to modify objects (for example, by cutting, welding, painting and cooking).
■ Collaboration: The ability to coordinate actions through cooperation with different things and to
combine actions to complete tasks such as multiagent assembly, lane merges and swarm
movements.
■ Autonomy: The ability to complete tasks with a minimum of external input, and to respond to a
dynamically changing space without recourse from cloud-based processing or other external
resources.
When exploring particular use cases for autonomous things, start with an understanding of the
space or spaces in which the thing will operate and the people, obstacles, terrain and other
autonomous objects it will need to interact with. For example, navigating a street is much easier
than a sidewalk because streets have lines, stoplights, signs and rules to follow. Next, consider the
outcomes you are trying to achieve with the autonomous thing. Finally, consider which technical
capabilities will be needed to address this defined scenario.
Combining Autonomous Capabilities and Human Control

Autonomous things operate along a spectrum of autonomy, from semiautonomous to fully
autonomous. The word “autonomous,” when used to describe autonomous things, is subject to
interpretation. When Gartner uses this term to describe autonomous things, we mean that these
things can operate unsupervised within a defined context or to complete a task. Autonomous things
may have various levels of autonomy. For example, a self-directing vacuum cleaner may have
limited autonomy and intelligence, while a drone might autonomously dodge obstacles.
It’s expedient to use the levels of autonomy often applied to evaluating autonomous vehicles when
39
considering use cases for any autonomous things:
■ No automation — Humans perform all the control tasks.

■ Human-assisted automation — Humans perform all the control tasks, but some control-assist
features are included in the design. The operator is still in charge of the thing, but specific
functions may be automated to simplify control.
■ Partial automation — Humans are responsible for control tasks and monitoring the
environment, but some control tasks are automated. The thing will be able to operate
autonomously for short periods, or within specific circumstances, but the operator is in constant
control.
■ Conditional automation — Control tasks are automated, but humans are required to take over
control tasks at any time at the request of the automation system. Autonomous operation is
possible, within certain operational bounds. The thing may be left to complete specific tasks,
but human intervention may be necessary at short notice.

■ High automation — Control tasks are automated under selected conditions, but humans are
required to take over control tasks when operating outside of those selected conditions. The
thing is designed to complete required tasks autonomously, and is capable of doing so, but may
require human intervention if circumstances change beyond specific bounds.
■ Full automation — Control tasks are automated under all conditions. Humans may request to
take over control tasks. The thing can complete all expected tasks without human intervention,
and is capable of adjusting the process to account for changing parameters to the point of
deciding when a task can only be partially completed.
Another consideration is the degree of direction or control that might be either required or desirable
regardless of the level of autonomy. Even with autonomous things that are fully autonomous and
can operate completely independently of any outside agency such as a human, it may be desirable
to insert a level of control or direction. In the case of an autonomous vehicle, for example, a human
would direct the destination and potentially other aspects of the driving experience and be able to
take control if desired. Consider autonomy and control as two complementary aspects to define
with regard to specific autonomous things use cases.
Autonomous, Intelligent and Collaborative

As autonomous things proliferate, we expect a shift from stand-alone intelligent things to a swarm
of collaborative intelligent things. In this model, multiple devices will work together, either
independently of people or with human input. For example, heterogeneous robots can operate in a
40
coordinated assembly process. In the delivery market, the most effective solution may be to use
an autonomous vehicle to move packages to the target area. Robots and drones aboard the vehicle
could then affect final delivery of the package. The U.S. Defense Advanced Research Projects
Agency (DARPA) is leading the way in this area and is studying the use of drone swarms to attack or
41
defend military targets. Other examples include:
■ 42
Intel’s use of a drone swarm for the opening ceremony of the Winter Olympic Games in 2018
■ SAFE SWARM by Honda, where vehicles communicate with infrastructure and one another to
43
predict hazards and optimize traffic flows
■ Boeing Loyal Wingman program deploys a series of drones that fly alongside a piloted vehicle
44
that can be controlled from plane it is escorting
Explore ways that AI-driven autonomous capabilities can power virtually any physical device in the
organization or the customer’s environment. Although autonomous things offer many exciting
possibilities, they cannot match the human brain’s breadth of intelligence and dynamic general-
purpose learning. Instead, they focus on well-scoped purposes, particularly for automating routine
human activities. Create business scenarios and customer journey maps to identify and explore the
opportunities that will deliver compelling business outcomes. Seek opportunities to incorporate the
use of intelligent things in traditional manual and semiautomated tasks. Examples of business
scenarios include:

■ Advanced agriculture. Projects such as the U.S. National Robotics Initiative are pushing
45
agriculture automation to the next level. Examples include creating planning algorithms for
robots to autonomously operate farms.
■ Safer automobile transportation. High-technology firms (such as Alphabet, Waymo, Tesla,
Uber, Lyft and Apple) and traditional automotive companies (such as Tesla, General Motors,
Mercedes-Benz, BMW, Nissan, Toyota and Ford) hope that, by removing the human error
element, self-driving cars will lower the number of automobile accidents. We anticipate more
46
than 1 million Level 3 and above cars will be produced annually by 2025.
■ Autonomous shipping. Maritime companies such as Kongsberg Maritime are experimenting
with various level of autonomous ships. Using an automated system built by Kongsberg, Fjord1
has automated two ferries operating between Anda and Lote in Norway. Kongsberg is also
47
working on autonomous container ships.
■ Search and rescue. A variety of autonomous things can operate to support search and rescue
in various environments. For example, “snakebots” made up of multiple connected and
redundant units can maneuver in tight spaces that are logistically impossible or unsafe for
people to explore. The Robotics Institute at Carnegie Mellon University is experimenting with a
48
variety of forms and locomotion models.
Related Research:
■ “Hype Cycle for Connected Vehicles and Smart Mobility, 2019”

■ “Hype Cycle for Drones and Mobile Robots, 2019”
■ “Toolkit: How to Select and Prioritize AI Use Cases Using Real Domain and Industry Examples”
■ “Swarms Will Help CIOs Scale Up Management for Digital Business”
■ “Use Scenarios to Plan for Autonomous Vehicle Adoption”
■ “Supply Chain Brief: Favorable Regulations Will Accelerate Global Adoption of Autonomous
Trucking”
Trend No. 9: Practical Blockchain

A blockchain is an expanding list of cryptographically signed, irrevocable transactional records
49
shared by all participants in a network. Each record contains a time stamp and reference links to
previous transactions. With this information, anyone with access rights can trace back a
transactional event, at any point in its history, belonging to any participant. A blockchain is one
architectural design of the broader concept of distributed ledgers. Blockchain and other distributed
ledger technologies provide trust in untrusted environments, eliminating the need for a trusted
central authority. Blockchain has become the common shorthand for a diverse collection of
distributed ledger products.
Blockchain potentially drives value in a number of different ways:

■ Blockchain removes business and technical friction by making the ledger independent of
individual applications and participants and replicating the ledger across a distributed network
to create an authoritative record of significant events. Everyone with permissioned access sees
the same information, and integration is simplified by having a single shared blockchain model.
■ Blockchain also enables a distributed trust architecture that allows parties that do not know or
inherently trust one another to create and exchange value using a diverse range of assets.
■ With the use of smart contracts as part of the blockchain, actions can be codified such that
changes in the blockchain trigger other actions.
Blockchain has the potential to reshape industries by enabling trust, providing transparency and
enabling value exchange across business ecosystems — potentially lowering costs, reducing
transaction settlement times and improving cash flow. Assets can be traced to their origin,
significantly reducing the opportunities for substitutions with counterfeit goods. Asset tracking also
has value in other areas, such as tracing food across a supply chain to more easily identify the
origin of contamination or tracking individual parts to assist in product recalls. Another area in which
blockchain has potential is identity management. Smart contracts can be programmed into the
blockchain where events can trigger actions; for example, payment is released when goods are
received.
According to the 2019 Gartner CIO Survey, 60% of CIOs expect some kind of blockchain
50
deployment in the next three years. In terms of industries that have already deployed blockchain or
plan to deploy it in the next 12 months, financial services leads the way (18%), followed by services
51
(17%) and transportation (16%). It is likely that organizations in these industries have a greater
need than organizations in other industries for the simpler use cases that limited use of some
blockchain components can support, such as record keeping and data management.
Blockchain Will Be Scalable by 2023

Blockchain remains immature for enterprise deployments due to a range of technical issues
including poor scalability and interoperability. Blockchain’s key revolutionary innovation is that it
eliminates all need for trust in any central or “permissioned” authority. It achieves that largely
through decentralized public consensus, which is not yet used in enterprise blockchain, where
organizations and consortia govern membership and participation. But enterprise blockchain is
proving to be a key pillar in digital transformation that supports evolutionary and incremental
improvements in trust and transparency across business ecosystems (see “Blockchain Unraveled:
Determining Its Suitability for Your Organization”).
By 2023, blockchain will be scalable technically, and will support trusted private transactions with
the necessary data confidentiality. These developments are being introduced in public blockchains
first. Over time, permissioned blockchains will integrate with public blockchains. They will start to
take advantage of these technology improvements, while supporting the membership, governance
and operating model requirements of permissioned blockchains (see “Hype Cycle for Blockchain
Technologies, 2019”).

Blockchain adds little value unless it is part of a network that exchanges information and value. The
network collaboration challenges have initially driven organizations to turn to consortia to derive the
most immediate value from blockchain. Choosing a consortium requires due diligence across
numerous risk criteria before sharing data and engaging with an outside party. Understanding and
evaluating these risks will be critical in extracting value from consortia involvement. Four types of
consortia exist: technology-centric; geographically centric; industry-centric and process-centric.
One option is to work with an industrywide consortium, but other types of consortia are also worth
exploring. Organizations need to carefully consider how these consortia will impact the enterprise
participation in particular industries and the competitive landscape.
Blockchain Use Cases

Gartner has identified multiple use cases for blockchain. These include:
1. Asset Tracking. These use cases cover the tracking of physical assets through the supply
chain to identify location and ownership accurately. Examples include tracking of automobiles
through loan processes, artworks postsale, and locations of ocean freight and spare parts.
2. Claims. This category covers automated claims processing in areas such as automobile,
agriculture, travel, and life and health insurance. It also includes other claims, such as
processing product recalls.
3. Identity Management/Know Your Client (KYC). This category covers uses where records
must be securely tied to an individual. Examples include managing records of educational
achievement, patient health, election identity and national identities.
4. Internal Record Keeping. In these use cases, the data to be secured remains within an
individual organization. Examples include master data management, internal document
management, purchase order and invoice records, and treasury record keeping.
5. Loyalty and Reward. This category includes use cases for tracking loyalty points (for retailers,
travel companies and others) and providing internal rewards, such as to employees or students.
6. Payment/Settlement. Use cases in this category involve a payment between parties, or
settlement of a trade. Examples include royalty payments, stock settlements, interbank
payments, commercial lending, procure-to-pay processing and remittance processing.
7. Provenance. Similar to the asset-tracking use case, this covers recording the movement of
assets, but the aim is to show the full history and ownership of the asset, rather than its
location. Examples include: tracking biological samples and organs; establishing the
provenance of wine, coffee, fish and other foods; certifying the authenticity of components; and
tracking pharmaceuticals through their life cycle.
8. Shared Record Keeping. This category includes use cases where data needs to be shared
securely between multiple participants. Examples include corporate announcements, multiparty
hotel booking management, recording of flight data and regulatory reporting.
9. Smart Cities/the IoT. This group includes use cases that use blockchain to provide data
tracking and to control functions for smart spaces or IoT solutions. These include peer-to-peer

energy trading, administration of electric vehicle charging, smart grid management and control
of wastewater systems.
10. Trade Finance. These use cases aim to streamline the process of financing trades, including
managing letters of credit, simplifying trade finance and facilitating cross-border trade.
11. Trading. Use cases in this group aim to improve the process for buying and selling assets,
including dealing in derivatives, trading of private equity and sports trading.
As this technology evolves, Gartner predicts that at least three other use cases will become more
viable and could prove the revolutionary benefits that blockchain enables:
■ Blockchain-based voting will benefit from improvements such as blockchain-related security,

management of forks, system governance and ledger interoperability. Blockchain could also
improve the tracking and traceability of voting tallies and voter rolls.
■ Blockchain-based, self-sovereign digital identity will also become more realistic. This could
rationalize the maze of the many-to-many suboptimal identity verification systems and
relationships used across the globe.
■ Cryptocurrency payments and remittance services will be used in countries with
hyperinflation, with cryptocurrency preserving purchasing power and financial well-being.
Despite the challenges, the significant potential for disruption and revenue generation means you
should begin evaluating blockchain, even if you don’t aggressively adopt the technologies in the
next few years. A practical approach to blockchain development demands:
■ A clear understanding of the business opportunity threat and potential industry impact
■ A clear understanding of the capabilities and limitations of blockchain technology
■ A reevaluation of your enterprise and industry trust architecture
■ The necessary vison and skills to implement the technology as part of core business strategy
■ A willingness and capability for customers to accept and adopt new operational constructs
Identify how the term “blockchain” is being used, both internally and by providers. Develop clear
definitions for internal discussions. Use caution when interacting with vendors that have ill-defined/
nonexistent blockchain offerings or thinly veiled repackaging of legacy offerings as blockchain
solutions.
52
Use Gartner’s spectrum model of blockchain evolution to better assess ledger developments,
including related initiatives, such as consensus mechanism development, sidechains and
blockchains. Resources permitting, consider distributed ledger as a proof of concept (POC)
development. But, before starting a distributed ledger project, ensure your team has the business
and cryptographic skills to understand what is and isn’t possible. Identify the integration points with
existing infrastructures to determine the necessary investments, and monitor the platform evolution
and maturation.
Related Research:

■ “Hype Cycle for Blockchain Technologies, 2019”
■ “Blockchain Technology Spectrum: A Gartner Theme Insight Report”
■ “The Future of Blockchain: 8 Scalability Hurdles to Enterprise Adoption”
■ “Use Gartner’s Blockchain Conceptual Model to Exploit the Full Range of Possibilities”
Trend No. 10: AI Security

Over the next five years AI, and especially ML, will be applied to augment human decision making
across a broad set of use cases. At the same time, there will be a massive increase in potential
points of attack with IoT, cloud computing, microservices and highly connected systems in smart
spaces. While this creates great opportunities to enable hyperautomation and leverage autonomous
things to deliver business transformation, it creates significant new challenges for the security team
and risk leaders. There are three key perspective to explore when considering how AI is impacting
the security space:
■ Protecting AI-powered systems. This requires securing AI training data, training pipelines and
ML models.
■ Leveraging AI to enhance security defense. This uses ML to understand patterns, uncover
attacks and automate aspects of the cybersecurity processes while augmenting the actions of
human security analysts.
■ Anticipating nefarious use of AI by attackers. Identifying these attacks and defending against
them will be an important addition to the cybersecurity role.
Protecting AI-Powered Systems

AI presents new attack surfaces and thus increases security risks. Just as security and risk
management leaders would scan their assets for vulnerabilities and apply patches to fix them,
application leaders must monitor ML algorithms and the data they ingest to determine whether
there are extant or potential corruption (“poisoning”) issues. If infected, data manipulation could be
used to compromise data-driven decisions that demand data quality, integrity, confidentiality and
privacy.
ML pipelines have five phases that need to be protected: data ingestion; preparation and labeling;
model training; inference validation; and production deployment. There are various types of risk
across each of these phases that organizations need to plan for. Through 2022, 30% of all AI
cyberattacks will leverage training-data poisoning, AI model theft or adversarial samples to attack
53
AI-powered systems.
■ Training-data poisoning: Hackers might have unauthorized access to training data and cause
an AI system to fail by feeding it incorrect or compromised data. Training-data poisoning is
more widespread in online learning models, which build and update their inputs as new data
comes in. ML systems trained on user-provided data are also susceptible to training-data
54
poisoning attacks, whereby a malicious user feeds bad training data with the goal of

corrupting the learning model. Reduce data-poisoning risk by limiting the amount of training
data each user contributes and examining output for shifts in predictions after each training
cycle.
■ Model theft: Competitors can reverse-engineer ML algorithms or implement their own AI
systems to use the output of your algorithms as training data. If competitors or bad actors can
observe data being fed into the AI and the output being reported, they can use this data to
55
develop their own ML models to engage in supervised learning and reconstruct the algorithm.
Researchers showed that certain deep learning algorithms are particularly vulnerable to such
55
imitation and manipulation. Detect model thefts by examining logs for unusual quantities of
queries or a higher diversity of queries, and protect the prediction machines by blocking
attackers and preparing a backup plan.
■ Adversarial samples: Classifiers are susceptible to a single sample of input/test data, which
can be altered slightly to cause an AI classifier to misclassify it. Most ML classifiers such as
linear classifiers (logistic regression, naive Bayes classifiers), support vector machines, decision
trees, boosted trees, random forest, neural networks and nearest neighbor are vulnerable to
56
adversarial samples. These alterations can be so refined that a human observer does not
notice the modification, yet the classifier still makes a mistake. Proactively defend against
adversarial samples by deploying a diverse set of prediction machines. Generate adversarial
samples and include them in your training dataset.
Leveraging AI to Enhance Cybersecurity Defense

As the rate and type of attacks expand, cybersecurity professionals will find it increasingly difficult
to keep up with demand for greater use of AI to filter and automate defense activities. Security tool
vendors are using ML to enhance their tools, decision support and response operations. With the
commoditization of research, frameworks and compute power, well-designed ML is within reach for
vendors with access to large quantities of relevant, high-quality training data. Assess solutions and
architecture, and challenge vendors on the latest ML-related attack techniques, including data
poisoning, adversarial inputs, generative adversarial networks and other security-relevant
innovations in ML.
ML-based security tools can be a powerful addition to your toolkit when aimed at a specific high-
value use case such as security monitoring, malware detection or network anomaly detection.
Supervised, unsupervised and reinforcement learning are successfully used in security today to
address malware, phishing, network anomalies, unauthorized access of sensitive data, user
behavior analytics, vulnerability prioritization and more. However, the best tools do not simply use
ML, but they deliver better outcomes. It is important to carefully evaluate and test vendors’ claims
57
to ensure they are delivering the value promised.
Well-designed security techniques based on ML are harder to

evade by attackers than rule-based techniques. However, the

probabilistic nature of ML can potentially generate many false
positives, make alerts more difficult to triage and increase the
complexity of tuning.
Generally speaking, using ML in defense forces attackers to change their techniques. Typically,
attacks become more complex and therefore more costly, but not impossible. In response, security
tool vendors and cybersecurity professionals develop new ML techniques to thwart these new
attack techniques. This long-standing back and forth will continue, and security teams need to be
prepared for the escalating tactics.
Attackers are innovating quickly to improve attacks on ML techniques used in security solutions.
They will attempt to poison training data and trick predictive algorithms into misclassification. Be
ready for attackers to adapt their techniques and evade detection by security techniques based on
ML. Do not rely on ML as a single prevention technique in security solutions.
ML-based security tools are not designed to fully replace existing traditional tools. Invest in ML-
based tools to augment security analysts. Ideally the ML-based tools will automate at one level
while freeing security analysts to focus on the more sophisticated and novel attack scenarios.
ML algorithms can also be used to improve tracking and cataloging data while providing more
accurate reporting and improving overall efforts to achieve compliance with privacy regulations such
as GDPR. The power of ML comes from algorithms that can identify patterns in and across
datasets. This capability can unearth otherwise hidden information that could drive a company out
of GDPR compliance. Chatbot technology and natural language processing can be used for
compliance dashboards, threat triage, security operations center (SOC) playbooks and reporting
automation. By automating the process of discovering and properly recording all types of data and
data relationships using AI applications, organizations can develop a comprehensive view of
compliance-related personal data tucked away in all structured and unstructured sources.
Developing a comprehensive view using traditional solutions and manual processes would be a
58
near impossible task.
Anticipating Nefarious Use of AI by Attackers

The next frontier of AI-related security concerns is emerging as attackers begin to use ML and other
AI techniques to power their attacks. Attackers have just started to leverage ML. They explore ML in
many security areas and are helped by the commoditization of ML tools and the availability of
training data.
Every new, exciting innovation in ML can and will be used

nefariously. Attackers will use ML to improve targeting, exploits,

the discovery of new vulnerabilities, the design of new payloads
and evasion.
Organizations must understand how adversaries may attack security solutions based on ML at
59
training and prediction stages, and how ML accelerates innovation in attacker methods.
Attackers leverage ML to accelerate innovation in attacker techniques. The commoditization of tools

for ML has led to many experiments on how to use these for nefarious purposes. One positive
industry development is that researchers have started to use some curation on AI publications and
60
code because of the potential use for malicious activity. Nevertheless, security professionals
should be aware of these developments and plan mitigations accordingly.
The first application for ML in attacks to explore is phishing. ML lends itself very well to digesting
large quantities of data, such as email or social media content. In phishing attacks, ML can be
useful for identifying targets, learning normal communication patterns, and then leveraging that
information to phish using the same communication styles. This works well for social media, since
there is often no lack of data that can be automatically parsed. In 2018, Cyxtera built an ML-based
phishing attack generator that trained on more than 100 million particularly effective historic attacks
61
to optimize and automatically generate effective scam links and emails. Cyxtera claims that, by
using AI based on open-source libraries and trained on public data, it was able to bypass an AI-
based detection system 15% of the time, whereas traditional approaches achieved this only 0.3%
of the time.
Identity deception is typically one aspect of phishing. The attacker’s intention is to gain the
recipient’s trust. The good news for attackers is that huge advancements in ML make identity
deception credible beyond what was expected a few years ago. ML can be used to simulate
someone’s voice in “vishing” attacks, or in voice authentication systems. ML can be used to fake
people in video. This form of nefarious ML use is commonly referred to as “deepfakes.” Nefarious
ML can learn what’s normal and adjust attacks within this “learned normal.” It can learn, simulate
and abuse writing styles, social graphs and communication patterns for deception.
Another recent example of nefarious use of ML is DeepExploit. DeepExploit uses Metasploit. It

trains on servers and finds the best payloads per target host (based on host info). Then it applies
fully automated intelligence gathering, exploitation, postexploitation and reporting. It leverages ML
(and signatures) to determine web products (intelligence gathering). Evidently, ML can determine the
used platforms/applications with more detail than signatures. DeepExploit then uses a pretrained
62
model to exploit this particular server.
Related Research:
■ “Anticipate Data Manipulation Security Risks to AI Pipelines”

■ “AI as a Target and Tool: An Attacker’s Perspective on ML”
■ “How to Prepare for and Respond to Business Disruptions After Aggressive Cyberattacks”

■ “How to Prepare for Cyber Warfare”
■ “Zero Trust Is an Initial Step on the Roadmap to CARTA”
Evidence
1 “Setting the ContinuousNext Foundation in Your Team”
2 “Seize the Technology Advantage With Combinatorial Digital Innovation”
3 “How to Build Segments and Personas for Multichannel Marketing”
4 “How Digital Products Drive Change in Product Design”
5 “Toolkit: Workshop for Creating EA Personas in Digital Business Diagnostic Deliverable Analysis”
6 “Business Events, Business Moments and Event Thinking in Digital Business”
7 “FromDigital Transformation to ContinuousNext: Key Insights From the 2018 Gartner Symposium/
ITxpo Keynote”
8 “Market Guide for Process Mining”
9 “Emerging Technology Analysis: Smart Wearables”
10 “Augmented Reality Is the Operating System of the Future. AR Cloud Is How We Get There,”
Forbes.
11 “3 Immersive Experience Use Cases That Provide Attractive Market Opportunities”
12 “Could You Fire Someone in Virtual Reality?” TWiT Tech Podcast Network. YouTube.
13 “Adopt a Mesh App and Service Architecture to Power Your Digital Business” and “API Mediation
Is the Key to Your Multiexperience Strategy”
14 “How to Use Machine Learning, Business Rules and Optimization in Decision Management”
15 “Drive Data Scientists’ Productivity With Data Preprocessing Techniques”
16 “Low-Code Development Technologies Evaluation Guide”
17 “Seek Diversity of People, Data and Algorithms to Keep AI Honest” and “Control Bias and
Eliminate Blind Spots in Machine Learning and Artificial Intelligence”
18 “What Is CRISPR Gene Editing, and How Does It Work?” The Conversation.

19 “Universal
Smart Contact Lenses Market 2018, Breathtaking CAGR of Approximately 10.4%,
Foreseeing 2023,” Medgadget.
20 “AI Helps Electronic Nose Distinguish Scents,” Daily Sabah.
21 “The Eyes Have It,” The Independent.
22 “Vagus Nerve Stimulator (VNS) Implantation for Children,” UPMC Children’s Hospital of
Pittsburgh.
23 “Military-Funded Study Successfully Tests ‘Prosthetic Memory’ Brain Implants,” Live Science.
24 “Speech Synthesis From Neural Decoding of Spoken Sentences,” Nature. (Paid subscription
required.)
25 “Elon Musk Hopes to Put a Computer Chip in Your Brain. Who Wants One?” CNN Digital.
26 “TheUntold Story of the ‘Circle of Trust’ Behind the World’s First Gene-Edited Babies,” American
Association for the Advancement of Science.
27 “Maverick* Research: Architecting Humans for Digital Transformation”
28 “Precautionary vs. Proactionary Principles,” Institute for Ethics and Emerging Technologies.
29 “Proactionary Principle,” Wikipedia.
30 “AI Is Sending People to Jail — and Getting It Wrong,” MIT Technology Review.
31 “Police Use Fitbit Data to Charge 90-Year-Old Man in Stepdaughter’s Killing,” The New York
Times.
32 “The FBI Used a Suspect’s Face to Unlock His iPhone in Ohio Case,” The Verge.
33 “The CIO’s Guide to Digital Ethics: Leading Your Enterprise in a Digital Society”
34 “Consumer Intelligence Series: Protect.me,” PwC.
35 “IEEE Global Initiative for Ethical Considerations in Artificial Intelligence (AI) and Autonomous
Systems (AS) Drives, Together With IEEE Societies, New Standards Projects; Releases New Report
on Prioritizing Human Well-Being,” IEEE Standards Association.
36 “Start Moving Data Management Capabilities Toward the Edge”
37 “RethinkYour Internal Private Cloud” and “When Private Cloud Infrastructure Isn’t Cloud, and
Why That’s Okay”
38 “2020 NASA Technology Taxonomy,” NASA.

39 “SAE International Releases Updated Visual Chart for Its ‘Levels of Driving Automation’ Standard
for Self-Driving Vehicles,” SAE International.
40 “Performance of Collaborative Robot Systems,” NIST.
41 “OFFensive Swarm-Enabled Tactics (OFFSET),” DARPA.
42 “Inside the Olympics Opening Ceremony World-Record Drone Show,” Wired.
43 “SAFE SWARM,” Honda.
44 “Boeing Unveils ‘Loyal Wingman’ Drone,” Defense News.
45 “National
Robotics Initiative 2.0: Ubiquitous Collaborative Robots (NRI-2.0),” National Science
Foundation (NSF).
46 “Forecast Analysis: Autonomous Vehicle Net Additions, Internet of Things, Worldwide”
47 “Norway’s Autonomous Ships Point to New Horizons,” Computer Weekly.
48 “Snake Robots Crawl to the Rescue, Part 1,” The American Society of Mechanical Engineers
(ASME).
49 D.
Furlonger, C. Uzureau. “The Real Business of Blockchain: How Leaders Can Create Value in a
New Digital Age.” Harvard Business Review Press. 15 October 2019.
50 See Figure 1 in “2019 CIO Agenda: Blockchain’s Emergence Depends on Use Cases.”
51 See Figure 5 in “2019 CIO Agenda: Blockchain’s Emergence Depends on Use Cases.”
52 “Understanding the Gartner Blockchain Spectrum and the Evolution of Technology Solutions”
53 “Anticipate Data Manipulation Security Risks to AI Pipelines”
54 “Certified Defenses for Data Poisoning Attacks,” arXiv.org, Cornell University.
55 “Stealing
Machine Learning Models via Prediction APIs,” SEC’16 Proceedings of the 25th
USENIX Conference on Security Symposium.
56 “Adversarial Vulnerability for Any Classifier,” arXiv.org, Cornell University.
57 “Assessing the Impact of Machine Learning on Security”
58 “Market Insight: Address GDPR Compliance With AI Applications”
59 “AI as a Target and Tool: An Attacker’s Perspective on ML”

60 The ML behind thispersondoesnotexist.com was not published because it was incredibly good at
generating realistic photos, and because of the power that such deception techniques can have
when used for nefarious purposes. Another example where results were not published because they
are too good and can be used nefariously is described in “The AI Text Generator That’s Too
Dangerous to Make Public,” Wired.
61 “AI Can Help Cybersecurity — If It Can Fight Through the Hype,” Wired.
62 Another
initiative worth mentioning in this regard is GyoiThon, another pentest tool that claims to
use ML. GyoiThon submits a large number of requests, not unlike DeepExploit, and claims to have
an ML engine (naive Bayes) that determines the application. The source code can be found at
GitHub.
Note 1 Strategic Technology Trends

Strategic technology trends create opportunities for incremental, transformative and even disruptive
innovation. Companies must examine the business impact of these trends and adjust business
models and operations appropriately or risk losing competitive advantage to those who do. These
are trends that IT cannot afford to ignore for a variety of reasons, such as:
■ High potential for impact/disruption

■ Reaching tipping points and/or rapidly evolving
■ Significant market activity and interest
■ Need to take a first/fresh look, next five years
■ Different from tactical (today), emerging (five to 10 years) and far horizons (more than 10 years)
Note 2 Robotic Process Automation

RPA learns about processes by recording keystrokes and mouse clicks associated with routine
labor related to structured applications and databases. RPA tools then mimic the activities and
decisions of human operations. The tools often cannot handle unstructured data, such as emails
and text. They are also brittle, and changes to the underlying applications (for example, field
changes and moved buttons) can break the scripted process.
Note 3 Intelligence at the Edge

The edge continues to grow in complexity. Intelligence will move to the edge across a spectrum of
endpoint devices. The line will blur between categories such as:
■ Simple embedded edge devices. This category includes switches, light bulbs, industrial
equipment, consumer electronics and appliances. Many existing devices in this category may
have little more than simple sensors and actuators. The trend is to drive more local compute
and storage, as well as more advanced sensors. In the consumer market, these simple devices
are morphing into more sophisticated devices. These include smart thermostats that can also
act as hubs for home automation systems and smart refrigerators that act in a similar way but

add more sophisticated sensors, local compute/storage and display technologies. The typical
life of consumer devices in this category is five to 20 years, while industrial assets often have a
life of 40 years. These long lead times create an environment with an enormous span of
capabilities in the installed base.
■ Edge I/O devices. This category includes devices such as speakers, cameras and screens. It
includes simple I/O-focused devices that rely heavily on capabilities delivered from local servers
and gateways or from external cloud services. Many of these devices will have greater local
capabilities, particularly in the form of embedded chips for AI and other specialized tasks. We
expect they will have a life of three to 10 years.
■ Edge computing devices. This category includes mobile devices, notebooks, PCs, printers
and scanners. These devices typically have a reasonable level of compute and storage function.
They will include an expanding array of sensors and more sophisticated AI chips in the next few
years. This category of device has a typical life of one to five years, so it has the greatest
potential for rapid updates. These devices will often act as localized processing or storage
systems for other edge devices in the environment.
■ Complex embedded edge devices. This category includes automobiles, tractors, ships,
drones and locomotives. Capabilities vary widely in the installed base but these devices will
increasingly become “mini-networks.” They will have extensive sensors, compute, storage and
embedded AI functionality with sophisticated communication mechanisms back to gateways,
servers and cloud-based services. These devices with have a life of three to 20 years, or even
longer. This, and the rapid expansion of advanced capabilities in new devices, will create
complexities for delivering a consistent set of functionality across a fleet of devices. This will
result in the delivery of as-a-service models in many sectors to promote more rapid refresh of
devices.
Note 4 The Elements of a Digital Twin

The essential elements of a digital twin are:
■ Model: The digital twin is a functional system model of the real-world object. The digital twin
includes the real-world object’s data structure, metadata and critical variables. More complex
composite digital twins can be assembled from simpler atomic digital twins.
■ Data: The digital twin’s data elements relating to the real-world object include: identity, time
series, current data, contextual data and events.
■ Uniqueness: The digital twin corresponds to a unique physical thing.
■ Ability to monitor: The digital twin can be used to query the state of the real-world object or
receive notifications (for example, based on an API) in coarse or granular detail.

GARTNER HEADQUARTERS
Corporate Headquarters
56 Top Gallant Road
Stamford, CT 06902-7700
USA
+1 203 964 0096
Regional Headquarters
AUSTRALIA
BRAZIL
JAPAN
UNITED KINGDOM
For a complete list of worldwide locations,

visit http://www.gartner.com/technology/about.jsp
© 2019 Gartner, Inc. and/or its affiliates. All rights reserved. Gartner is a registered trademark of Gartner, Inc. and its affiliates. This
publication may not be reproduced or distributed in any form without Gartner's prior written permission. It consists of the opinions of
Gartner's research organization, which should not be construed as statements of fact. While the information contained in this publication
has been obtained from sources believed to be reliable, Gartner disclaims all warranties as to the accuracy, completeness or adequacy of
such information. Although Gartner research may address legal and financial issues, Gartner does not provide legal or investment advice
and its research should not be construed or used as such. Your access and use of this publication are governed by Gartner Usage Policy.
Gartner prides itself on its reputation for independence and objectivity. Its research is produced independently by its research
organization without input or influence from any third party. For further information, see "Guiding Principles on Independence and
Objectivity."

Top 10 Strategic Technology Trends For 2020

Uploaded by

Copyright:

Available Formats

Top 10 Strategic Technology Trends For 2020

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Top 10 Strategic Technology Trends For 2020

Uploaded by

Copyright:

Available Formats

What are the top 10 strategic technology trends for 2020 according to the report?

What are the top 10 strategic technology trends for 2020 according to the report?

What are some recommendations for enterprise architecture and technology innovation leaders according to the report?

What are some recommendations for enterprise architecture and technology innovation leaders according to the report?

Top 10 Strategic Technology Trends for 2020

Published: 21 October 2019 ID: G00432920

Strategic technology trends have the potential both to create opportunity

Strategic Planning Assumptions............................................................................................................. 3

Page 2 of 52 Gartner, Inc. | G00432920

Figure 1. Top 10 Strategic Technology Trends for 2020...........................................................................5

Strategic Planning Assumptions

By 2021, at least one-third of enterprises will have deployed a multiexperience development

Gartner, Inc. | G00432920 Page 3 of 52

By 2020, we expect that 4% of network-based mobile communications service providers (CSPs)

Page 4 of 52 Gartner, Inc. | G00432920

Gartner, Inc. | G00432920 Page 5 of 52

People-Centric Smart Spaces Build on the Intelligent Digital Mesh

Page 6 of 52 Gartner, Inc. | G00432920

Gartner, Inc. | G00432920 Page 7 of 52

Page 8 of 52 Gartner, Inc. | G00432920

Gartner, Inc. | G00432920 Page 9 of 52

RPA and iBPMS Are Key Components of Hyperautomation

Page 10 of 52 Gartner, Inc. | G00432920

The Digital Twin of an Organization

Gartner, Inc. | G00432920 Page 11 of 52

Machine Learning and NLP Explode the Range of Hyperautomation Possibilities

Page 12 of 52 Gartner, Inc. | G00432920

Trend No. 2: Multiexperience

The model will shift from one of technology-literate people to

Gartner, Inc. | G00432920 Page 13 of 52

VR provides a 3D environment that surrounds a user and responds to an individual’s actions in a

Page 14 of 52 Gartner, Inc. | G00432920

The Language Factor

Gartner, Inc. | G00432920 Page 15 of 52

The Multiexperience Development Platform

MXDPs are not “build once, run everywhere” or “omnichannel” solutions.

The core value of an MXDP lies in its ability to coalesce

Page 16 of 52 Gartner, Inc. | G00432920

■ “Market Guide for Conversational Platforms”

Trend No. 3: Democratization

■ Democratization of Application Development. AI PaaS provides access to sophisticated AI

Gartner, Inc. | G00432920 Page 17 of 52

Democratization of Application Development

Not only does the evolution of these AI platforms and suites of

Page 18 of 52 Gartner, Inc. | G00432920

Another level of democratization occurs as AI is applied to automate various application

Ultimately, highly advanced AI-powered development

Augmented Analytics and the Citizen Data Scientist

Gartner, Inc. | G00432920 Page 19 of 52

Dealing With Shadow AI

Page 20 of 52 Gartner, Inc. | G00432920

■ “Predicts 2019: The Democratization of AI”

Trend No. 4: Human Augmentation

Gartner, Inc. | G00432920 Page 21 of 52

Cognitive and Physical Augmentation

Physical augmentation enhances humans by changing their inherent physical capabilities by

■ Sensory Augmentation — Hearing, vision and other augmentation devices or implants to

Page 22 of 52 Gartner, Inc. | G00432920