Artificial Intelligence

The 3 R’s of AI
Adoption
Refactor.
Reinvent.
Reimagine.
by Kimberly Nevala

best
practices
THOUGHT PROVOKING BUSINESS
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

Table of Contents

Introduction........................................................................................ 3

The Devil is in the Definition............................................................ 3

Three Levels of AI Adoption............................................................ 4

Assessing Impact............................................................................... 5

Level 1: Refactor................................................................................. 6

Level 2: Reinvent................................................................................ 9

Level 3: Reimagine..........................................................................12

AI across The Spectrum..................................................................15

Climbing the AI Ladder..................................................................21

2
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

INTRODUCTION
Artificial Intelligence (AI) is everywhere – or so it seems. Examples of
AI’s practical potential are rampant and increasing seemingly without
bound: the possibilities are equally exciting and overwhelming. Rapidly
progressing and increasingly available AI capabilities provide opportu-
nity while the sheer scope of possibility and impact on existing business
relationships and practices is intimidating. This can make it difficult to
know where to start, especially when our natural inclination is to “go big
or go home.” But do you have to upset the proverbial apple cart to even
get started with AI? Not necessarily!

Read on to explore three levels at which organizations are adopting AI

to drive business forward: refactoring, reinventing and reimagining. The
risks and rewards inherent in each as well as the relative impact of AI on
existing processes, data and IT practices are examined. While the relative
scope of change increases dramatically moving up the chain, each level Analytically
provides germane opportunities for business improvement and quantifi-
able business value. In the end, analytically savvy organizations will apply
savvy
AI across all three vectors. Where you start depends strictly on your orga- organizations
nization’s goals, analytics maturity and ability to execute.
will apply
AI across all
three vectors.
THE DEVIL IS IN THE DEFINITION
Where you
Perceptions of AI vary: from visions of autonomous, independent human- start depends
oids to narrow definitions focusing on the ability to automate tasks previ-
ously thought to be the sole purview of humans.
strictly on your
organization’s
Practically speaking there is general agreement that AI solutions (aka:
algorithms) are those that are able to learn from, sense and intelligently
goals, analytics
respond to events (aka: data) in their environment. More specifically, SAS maturity and
defines AI as the science of training computers to emulate human tasks
through learning and automation. With AI machines can learn from expe-
ability to
rience, adjust to new inputs and accomplish specific tasks without manual execute.
intervention.

3
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

The ability to learn from input data without being explicitly programmed
(i.e., told the rules in advance) and to interact with other parties (mostly us
humans) in a natural and intuitive manner are hallmarks of AI. To that end,
a number of technologies underpin AI solutions including machine and
deep learning, natural language processing (NLP) and computer vision.
Which elements are brought to bear depends on the problem being
solved. As does the answer to the questions: “do I need AI?” and “what
can AI do for me?”

THREE LEVELS OF AI ADOPTION

While some AI applications are spellbinding in the audacity of their vision
and scope most are more mundane, yet no less impactful. As shown in
Figure 1, there are three approaches to consider when identifying oppor-
tunities for AI.

FIGURE 1
The Three Levels
of AI Adoption

AI Maturity

REFACTOR – Apply AI to enhance system behavior and performance

within the context of existing business processes. Refactoring changes
the internals of business processes and systems without affecting their
external behavior.

4
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

REINVENT – Apply AI to modify how and when processes and services

are delivered. Reinvention changes how things work in ways that appear
to be different and new.

REIMAGINE – Apply AI to be disruptive. Reimagining changes the playing

field entirely: deploying products, services and engagement models
never seen before.

ASSESSING IMPACT
Regardless of where you start, implementing AI will require changes to
existing business and technical practices. The scope of the change varies
drastically as one moves from refactoring existing processes to reimaging
net new approaches. When assessing the organization’s willingness and
ability to implement a proposed AI solution, consider these six dimen-
sions:

• Business Process – Extent to which existing product/service of-

ferings, business practices, operating models, employee roles
and skills are affected.
Regardless
• Analytics Maturity – Level of analytic maturity, specialized solu-
tions and skills required to execute. of where
• D
 ata Enablement – Sophistication and scope of required data you start,
practices from governance to management including consider- implementing
ations for data availability, quality, privacy and usage.
AI will require
•  evOps – Implications for traditional IT practices including stan-
D
dard operating practices for deployment, ongoing monitoring changes
and maintenance of AI algorithms and models. to existing
• Risk – Overarching risk inclusive of organization’s ability to man- business and
age change; political or regulatory exposure; implications for
external relationships (customer, partners and so on). technical
• V
 alue/ROI – Value of delivered business outcome and return on
practices.
investment.

5
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

LEVEL 1: REFACTOR

What It Is
The objective of refactoring is to optimize existing processes: in layman’s
terms, the intent is to perform existing jobs better and faster. When AI is
applied in this context, the external behavior of the process or system is
unaffected. To all outward appearances – from the customer’s perspec-
tive, for example – there is little to no observable difference in how they
engage with the organization. Behind the scenes, however, AI may be
applied to automate high-volume, repetitive decision points and actions.
At this level, AI can also be applied to supercharge existing analytics
models through incorporation of previously unexplored data sources or
application of new modeling techniques.

Refactoring:

•  ocuses primarily on making existing practices more effective

F
and efficient.

• Does not change existing product/service portfolio.

• Improves how but not what or when products and services are The objective
delivered.
of refactoring
• Automates rote, high-volume decision points and tasks.
is to optimize
• Delivers incremental improvements.
existing
Using AI to refactor existing practices may seem underwhelming, if not processes: in
downright anticlimactic, given the public hype around AI’s transforma-
tive potential. While understandable, organizations early in the AI journey layman’s terms,
should not shirk from refactoring. Refactoring projects may not be the the intent is to
sexiest on the surface. However, such projects provide a great jump off
point to incrementally introduce and hone new skills and analytic methods perform existing
required for more ambitious applications. Refactoring also provides a jobs better and
great opportunity to prove the value of AI in a business context that is
already well understood and where expected outcomes can be clearly faster.
measured. This substantially lowers overall risk and barriers to entry.

6
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

What It Takes
Refactoring provides the opportunity for organizations to
begin to introduce new analytics and data practices required
to support a robust AI practice incrementally.

What It Looks Like

Business Process

DEFECT DETECTION
IN MANUFACTURING
While some routine tasks may be automated and existing
analytics given an IQ boost, the overall business process flow Utilizing text analytics to
and participants are largely unaffected. Parties external to incorporate text previously
the process (customers, partners, etc.) are unlikely to notice held hostage within free form
a change in how the organization engages with them. They
call center notes and service
may notice that services or processes are more efficient.
technician logs to enhance
predictive models used to
Analytic / AI Maturity detect manufacturing defects
early in the production cycle.

SERVICE ROUTING
Existing analytics practices are extended by adding one
additional element or dimension. For instance, adding text
IN INSURANCE
analytics to interrogate written text contained in pre-existing
dark (previously unexplored) data sources such as service Applying natural language
logs. processing and voice analysis
to detect stress and customer
Data Practices sentiment in order to appro-
priately prioritize and route
customers calling to report
an accident or file a claim.
May result in acquisition of a single new data source. More
often, refactoring projects integrate existing sources in
new ways and/or extending data management practices to
accommodate previously unmanaged data types such as
text.

7
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

DevOps

Existing analytics models may be enhanced or replaced with more sophis-

ticated algorithms but not deployed differently. Therefore, overall impact
on IT practices created to support more traditional reporting applications
and less sophisticated analytics is limited.

Risk

Change management is largely constrained to analytics development

and deployment. Automation of rote business activities can result in
minor adjustments to internal practices and may disrupt jobs of some
low- to medium-skilled resources as discrete activities are offloaded to
the AI solution. On balance, external parties including employees and
customers are unlikely to notice major changes in the way they work.

Realized Value/ROI

Refactoring provides a path to slowly introduce new technology and Refactoring

build the business case for further investment in broader or more specu-
lative projects. Automation, in particular, delivers cost reductions due to
may also impact
increased operational efficiencies. Refactoring may also impact revenue revenue or
or satisfaction metrics through improved process effectiveness.
satisfaction
metrics through
improved
process
effectiveness.

8
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

LEVEL 2: REINVENT

What It Is
Organizations looking to reinvent themselves apply AI to improve how
core services and products are created and delivered. Reinvention
changes how things work in ways that make the product or service appear
to be different and new. At this level, the external behavior of the system/
interface is impacted. However, the portfolio of products/services offered
remains consistent. Key functions or tasks with a higher level of predict-
able variability are now offloaded to the AI solution allowing human coun-
terparts to concentrate on higher-value tasks or interactions. At this stage
AI begins to play a role as an advisor, augmenting the abilities of human
decision makers and workers by providing targeted insights and advice.

Reinvention:

•  hifts focus from automation to include

S
augmentation.

• Integrates AI into existing products and services.

•  hanges how and when customers, partners and employees

C Organizations
engage.
looking to
• O
 ffloads discrete functions and tasks to AI as a
visible actor in the process. reinvent
• May introduce new revenue opportunities. themselves
apply AI to
It is at this point that the AI solution becomes a visible entity within and
outside of the enterprise. Employee roles and responsibilities morph as improve how
business processes are functionally reengineered to fully exploit more core services
advanced AI capabilities. Customers and partners may notice that how
and when they engage with an organization is changing. As a result, the and products
AI user experience becomes important: relationships, interaction points are created and
and handoffs between AI systems and those they interact with must be
carefully designed. In addition, organizations must carefully assess both delivered.
real and perceived risks as parties adjust to a new method of working with
each other and the “machine” (aka: the AI solution).

9
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

What It Takes

Business Process
What It Looks Like
CUSTOMER
Existing tasks or responsibilities may be executed by AI or
SERVICE IN
processes changed to incorporate improved insights. In FINANCIAL SERVICES
both cases, existing human roles are affected requiring ei-
ther new skills or learning to work side-by-side with the “ma- Utilizing a chat bot to handle
chine” (so to speak). Business development is also key as routine customer inquiries and
how customers and other parties engage with the organiza- facilitate basic self-service func-
tion is affected. tions like requesting a state-
ment or a new card, resetting a
Analytic / AI Maturity password, transferring money
between accounts and so on.

DIFFERENTIAL
Extending traditional advanced analytics practices with AI DIAGNOSIS IN
capabilities requires acquisition of resources experienced HEALTHCARE
in AI/ML development.
Applying advanced image
Data Practices processing and machine
learning to identify potential
malignancies on medical
images and provide this input
Requires robust production data chains that break down to MD for use in diagnosis
existing barriers imposed by traditional data architec-
and treatment planning.
tures that strictly delineate operational environments from
analytics.
Analyzing data from medical
monitors/sensors in real time to
identify an emergent infection
in an infant in the PICU before
clinical symptoms are observed.

10
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

DevOps

AI algorithms require continuous monitoring and tuning to ensure busi-

ness results are delivered, and the solution appropriately responds to
behaviors or conditions not originally present. It is at this point that tradi-
tional IT delineations between development and operations become
fraught. Just as business roles must change, so too shall existing DevOps
processes and SLA.

Risk

While there isn’t a wholesale rethink of what is sold, existing roles must be
modified as outward-facing functions migrate to AI and to allow AI to func-
tion as an effective collaborator with a human counterpart. If deploying
AI in an external facing role – in the case of a chatbot or personal assistant,
for example – a keen understanding of how the reactions or expectations
of the customer/participant may change when a “machine” is involved is
critical. A keen
understanding
Realized Value/ROI – Medium
of how the
reactions or
Delivers both cost savings and incremental revenue generation. Increased
expectations of
engagement and retention from your customers/partners or employees the customer/
can also occur.
participant may
change when
a “machine”
is involved is
critical.

11
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

LEVEL 3: REIMAGINE

What It Is
Reimagining is the proverbial “greenfield” in which the very nature of the
industry, product or service undergoes a wholesale rethink. Undertak-
ings at this level are often heralded as ‘AI-first’ or ‘AI-driven’ undertakings.
While this may overstate the case for the sake of marketing and inveigling
investment, at this stage AI is an integral part of the delivered product/
service. In other words, the product or service couldn’t be delivered
without AI. Additionally, business processes and systems are engineered
with the AI solution as an active collaborator (actor) in the process.

Reimagining:

• Focuses on rethinking the playing field.

• Introduces new products, services and engagement models.

• AI is integral to product/service design and delivery.

• Introduces new roles and responsibilities.

• Creates new revenue streams. Reimagining is

the proverbial
At this juncture, mindful and seamless integration between human and
machine becomes critical. Pre-existing business functions and roles are at “greenfield”
a minimum redefined or may become obsolete. Net new jobs are often in which the
created as new engagement models are developed. Analytics and data
ecosystems are highly mature, and DevOps practices have evolved to very nature of
support the continuous improvement mindset required to maintain algo- the industry,
rithms in the face of continuous environmental change.
product
or service
undergoes
a wholesale
rethink.

12
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

What It Takes
At this level, AI’s impact is felt on every level.

Business Process
What It Looks Like
THE SELF-
Net new business and engagement models are being
DRIVING CAR
defined. Older business models and systems may be retired
as new services, products or engagement models are Deploying sophisticated solu-
deployed. Acquisition of resources with specialized skills tions incorporating machine
and retraining of existing resources to accommodate new and deep learning, object
ways of working are commonplace. identification and detection
(visual processing) to allow a
Analytic / AI Maturity car to navigate the most opti-
mal route from point A to point
B without human intervention.

Very sophisticated algorithms are utilized and integrated THE CASHIER-

as fundamental components of operational systems. Auto- LESS STORE
mation and augmentation are both prevalent. Multiple
AI systems work in concert to support end-to-end busi- Using facial recognition, ML and
ness processes within the internal and external business
NLP to identify and greet a cus-
ecosystem.
tomer upon entry to the store.
Subsequently detect, identify
Data Practices and maintain rolling account
of items in the cart and auto-
bill customer upon exit. Along
Requires a high degree of sophistication. At this level, data
the way, remind the shopper
engineering is fully recognized as a discrete, critical func- of additional items they may
tion and data considerations are an inherent component be forgetting based on past
of business process and system design. Because business shopping patterns, season,
processes and systems are being invented, requisite data preferences and current deals.
may not be captured by existing processes or is only avail-
able to a select few initially.

13
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

DevOps

DevOps at this level is very sophisticated. Standard operating practices

support a continuous modeling mindset in which algorithmic mainte-
nance and tuning occurs on an as-needed, right-in-time basis. In addi-
tion, AI deployment is increasingly modular - creating agility by enabling
access and mix-and-match reuse of algorithms to support multiple use
cases across platforms.

Risk

New business models are – by definition – speculative. Because these very

often challenge long-standing operating models and perceptions, both
concrete and perceived risks, including ethical considerations and impact
on jobs, are front and center. Existing public policies and regulatory/com-
pliance may not accommodate new modes, requiring substantial partner-
ships and advocacy to bring the proposed product/service to fruition.

Realized Value/ROI
Both concrete
and perceived
risks, including
Creates net new revenue streams. New business models are most likely
to increase engagement of existing customer and engage a new cadre of ethical
users previously unknown to the organization. While created value may
considerations
be slightly offset by losses or slowing returns in traditional offerings, the
long-term ROI potential is highest at this stage. and impact on
jobs, are front
and center.

14
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

AI ACROSS THE SPECTRUM

Healthcare

Incorporate patient sentiment, sociographic and behavioral

Predict Compliance data into risk profiles to analyze and identify patients likely to be
uncompliant with medication or physical therapy prescriptions.

Apply natural language processing and machine learning to

Auto-Coding
identify and correct common diagnostic and billing coding errors
& Correction
or discrepancies based on individual patterns.

Surgical/Medical Unit Optimization; Patient Flow Forecasting;

Other Medical Fraud Detection (prescriber; provider; patient);
Population Health Monitoring; Provider Quality Analysis

Utilize deep learning to predict patient prognosis based on

Predict Patient
comprehensive analysis of published research, clinical, claims,
Prognosis
and environmental factors.

Identify potentially efficacious drugs based on analysis of

Drug Discovery genome and protein structures coupled with identification and
analysis of clinical data and outcomes for “like” illnesses.

Disease Detection; Differential Diagnosis; Trial Design &

Other Optimization; Bio-Surveillance: Detect & Preempt Disease
Outbreaks; Supply Chain Optimization

Utilize data from wearables to identify motor-disease based

on a patient’s pattern of movement or sensory data. Apply
The Connected
same information to monitor disease progression, measure the
Patient
progress of rehabilitation or to manage pain in real time based
on collected data, not subjective patient reporting.

Determine optimal diet, medication or treatment based on

evaluation of a patient’s genetic profile, microbiome, clinical
Precision Medicine and behavioral records. Analyze discrete morphological
and molecular changes to monitor and optimize targeted
interventions for each patient.

Virtual Personal Medical Assistant, Artificial Limb Animation;

Other
Robotic Surgery

15
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

Manufacturing

Apply image processing and analytics to streaming image data

to identify defects in silicon wafers invisible to the human eye
Defect Detection
and flag for operator to remove before the product leaves the
production line.

Apply visual image processing and ML to allow production robots

Self-learning
to learn and optimize a discrete function such as spot welding
Robots
without human invention.

Quality Assurance and Control; Production Optimization; Self/

Other
Spatially Aware Robots

Predict when a product such as a wind turbine or engine will

fail or malfunction based not just on engineering specifications
Predictive and average historical performance data but operating history,
Maintenance current performance and environmental conditions for the
specific device. Alert to enable corrective action just in time,
before outage or critical failure occurs.

Incorporate real-time traffic, weather and other operating data

to optimize delivery times and minimize fuel consumption in
Fleet Optimization
real time. Incorporate ongoing learnings into future planning to
optimize fleet configuration and personnel.

VR-Enabled Service; Collaborative Robotics; Asset Performance

Other
Optimization; Predictive Inventory Management

Systematically identify and remove defective materials and

adjust operating parameters to resolve errors, optimize asset
performance and maximize throughput. Automatically and
Smart Factory
continuously adapt manufacturing output in response to
fluctuations in supply chains, distribution networks and consumer
demand.

Apply reinforcement learning to design product based on

specified goals and objectives, analysis of known engineering
Product Design/
and science principles, prior product specifications, known
Engineeering
design patterns and so on. Discuss constraints and iterate with
human designer.

Self-Driving Vehicles; Smart Products (IOT-enabled); Mass

Other
Customization

16
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

Banking/Insurance

Utilize AI to automate repetitive tasks such as authenticating and

Robotic Process
correcting errors in fund transfer requests, reconciling failed
Automation (RPA)
trades, and responding to routine data requests.

Apply semi-supervised and unsupervised machine learning

AML/Fraud techniques to detect potential fraud through analysis of known
Detection patterns and detection of new anomalous transaction or network
patterns by and between individual parties/entities.

Cybersecurity and Network Intrusion Detection; Biometric

Other Identify Verification (facial recognition); Stress Testing; Document
Classification and Analysis

Utilize natural language processing and ML to understand a

written or spoken request, take appropriate action or provide
Virtual Customer
requested information to customer in their native tongue.
Service Agent
Reroute non-standard inquiries or disaffected customers with
contact notes to human agent for follow-up.

Apply natural language processing and machine learning to

determine probable resolution and cost of new claim based
Automated Claim on client’s policy, profile, claimant behavior and claim history.
Resolution Authorize repair or replacement for low-value or low-risk claims
automatically, route complicated and potentially fraudulent
transactions to human agents/investigators.

Predictive Default Detection; Predictive Client Outreach/Financial

Other
Health Monitoring; Distracted Driver Detection and Alerts

Continuously identify potential assets and execute investment

recommendations (i.e. buy stock, adjust distributions) based on
Investment individual financial goals, principles (e.g. ecofriendly, charitable)
Optimization and ongoing evaluation of emerging business news, corporate
reports/prospectus, global and local consumer, geosocial and
political trends.

Utilize AI to monitor, adjust and optimize a customer’s financial

portfolio across asset /account types and service providers (e.g.
Adaptive
bank, investments and insurance) based on evolving needs and
Portfolio/Wealth
goals determined not just by stated objectives but detection
Management
of major life events, current behaviors, market and consumer
signals.

17
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

Retail

Apply natural language processing and ML to analyze the

content and tone (i.e. sentiment) of a customer calling or writing
Customer Service
into a contact center or posting on social media to inform
Optimization
appropriate outreach and identify emerging problem areas and
consumer trends sooner.

Apply image processing and text analytics to identify, categorize

Product Content and relate information from product images, specifications,
Management consumer commentary to properly identify, catalogue and cross-
reference content (i.e. identify “like” styles).

Utilize natural language processing to enable a virtual concierge

to engage a customer in series of contextual questions (in writing
Virtual Concierge or verbally). Recommend the appropriate product or service by
taking into account individual answers and consumer trends,
buying behaviors and preferences of like customers.

Utilize machine learning to integrate, analyze and recommend

Contextual products/services based on individual profile, buying history,
Marketing/Hyper- current location and activity. Automatically engage customer
personalization through appropriate channel using natural language interfaces
and/or traditional outreach mechanisms.

Dynamic Pricing; Smart Service Routing; Inventory Optimization;

Other
Supply Chain Optimization

Deploy a virtual personal shopper to identify clothing needs

based on inventory and visual assessment of current wardrobe
(styles, colors), event calendars, season, historical buying
Personal Shopper patterns and/or explicit requests. Identify likely designers/
designs, present individual mockups, shop for best outlet –
retail or bespoke and execute purchase automatically or upon
approval.

Predict short-term or real-time forecasts for demand based on

evaluation of sales, market product trends, current economic
Supply Chain
and geopolitical climates, consumer sentiment and interest as
Optimization
expressed on social networks and so on. Automatically adjust
procurement, sales, pricing and fulfillment models accordingly.

Cashier-Less Store; Predictive Fulfillment;

Other
Personal Product Design

18
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

Energy/Utilities

Apply natural language and image processing to identify

and catalogue disparate documents (contracts, land deeds,
Content regulatory policies, asset specifications, etc.). Perform contextual
Classification & analysis to identify and highlight key terms, sentiment (if
Analysis applicable), and requirements within and between different
policies, contracts and specifications for action by human
knowledge worker.

Apply deep learning to optimize energy consumption forecasts

Energy Forecasting by accounting for both short- and long-term variabilities in
operating conditions and resultant long-term dependencies.

Outage Prediction & Detection; Cybersecurity; Disaster

Other
Monitoring & Response

Analyze environment-specific usage patterns and preferences to

Connected Home optimize comfort while minimizing usage and cost. Detect and
[smart thermostat] adjust heating and cooling based on factors such as the number
of people and level of activity present.

Optimize HVAC and related control systems in data centers

Consumption against acceptable operating parameters, variable and hard-to-
Management predict environmental factors and demand conditions, in real
time.

Asset Performance Management; Smart Meters; Portfolio

Other
Optimization; Customer Personalization/Target Marketing

Deploy integrated AI solutions to seamlessly and autonomously

adjust energy generation and delivery between disparate
The Smart Grid producers, energy sources, and consumers to maximize use of
renewable or low-cost energy sources, optimize energy mix, and
avoid outages.

19
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

Other Cool Applications

Employ a virtual executive assistant to answer rote emails and

inquiries using your personal tone of voice. May also automate
basic workflows (prioritizing emails, accepting calendar invites
Executive Assistant
or approving time off and expenses) and collate information
required for key decisions or meetings in an integrated format
with initial analysis to improve response time.

Apply machine learning and natural language processing to

perform legal research or evaluate contracts by identifying,
Virtual Paralegal collating and analyzing relevant content including applicable
laws, regulations, case law, precedents, as well as public opinion
and sentiment via social media and news publications.

Utilize image processing, object detection and deep learning

to track the identify and movement of all the soccer players on
Sports Performance a field to inform coaching and game day strategies, identify
Optimization highly-performing and high-potential players (and vice versa)
and optimize overall team performance across diverse playing
conditions and scenarios.

Identify and track endangered species in remote and

Non-Invasive inaccessible areas on an ongoing basis by applying image
Wildlife processing to the footprints they leave behind. Utilize ML to
Conservation analyze and predict population health, stability and emerging
threats or needs for investigation.

Apply data from sensors, imaging, social media to adjust traffic

patterns, street lighting, parking, and optimize deployment of
Smart City services such as transportation, waste management, police or
public health in real time. Also used to improve forecasting,
planning and budgeting for critical public resources.

20
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

CLIMBING THE AI LADDER

Is This Really New?

You likely noticed that many of the applications showcased above are not
net new. Personalized marketing, for example, has been near and dear
to retailer’s hearts for decades. Literally, decades. During which time,
progressively maturing analytic capabilities have played an increasingly
important role in realizing the goal of true personalization.

So what makes AI different? First, the way AI algorithms are developed

(i.e., learn) as well as the diversity (video, images, text, IOT) and volume
of input they can ingest has vastly changed and improved. Beyond this,
there are two key determinants for what tends to be publically* recog-
nized as AI. These two AI hallmarks are the ability to:

•  eploy the system to sense-and-respond to events in a contex-

D
tually appropriate way IRL (in real life). Which doesn’t always,
but very often, translates to real-time operation.

• Interact with the system using natural language – interpreting, AI provides a

speaking and writing in “plain English” (or your primary lan- mechanism
guage du jour).
to improve
The questions that AI answer are not fundamentally different than previous
generations of analytic solutions: what happened, why, what will happen and extend
next, what should we do about it? Rather, AI provides a mechanism to the scope,
improve and extend the scope, effectiveness and usability of previous
analytics applications. This, in turn, creates opportunities for new engage- effectiveness
ment and business models. and usability
That said, while it’s important to have clarity about the term AI, it is more of previous
critical to have clarity about the problem being solved or objective at analytics
hand. Despite its emergence on marketing pages everywhere, not every
problem requires AI. Nor can AI solve any problem. applications.

* As opposed to more formal, perhaps rigorous, and philosophical arguments about
the nature of intelligence by academics, researchers and other curious sorts like I. 21
The 3 R’s of AI Adoption: Refactor, Reinvent, Reimagine Artificial
Intelligence

The Power of the Collective

It is also noteworthy that the most complicated and immersive AI solu-
tions are made up of a network of smaller solutions working in concert.
Comprehensive AI environments function much like an orchestra in which
individual players contribute uniquely and vitally to the outcome of the
whole. The result of which, is of course, more complicated and awe-
inspiring (or, at a minimum, louder) than any individual could achieve
alone. To that end, solving extremely complex problems with AI starts by
identifying a series of discrete decision points or actions. Followed by
integrating those solutions into an ensemble. This is an important point
because it illuminates a path forward for organizations that do not yet
have the analytic, data and organizational willpower or experience to
deploy AI pervasively.

Where Will You Begin?

Contrary to often sensationalized accounts touted on the evening news,
AI – like other game-changing technologies before it – can be applied to
your business in big and small ways. To remain competitive, the ability
to apply AI to refactor and reinvent core business practices may be a de
facto requirement in the near future. For the time being, reimagining,
while the most tantalizing, will likely remain the exclusive domain of
innovative upstarts, organizations that are “born digital” and particularly AI – like
forward thinking organizations with the money, talent and willpower to
fundamentally rethink how the world works.
other game
changing
Regardless, there is value to be had. It’s certainly tempting to run straight
for that greenfield and reimagine a different future for your business.
technologies
However, this will be an exercise in frustration if the organization does before it – can
not have the analytic and data muscle, along with both the willingness
and ability to make broad changes to existing business practices. Don’t
be applied to
despair if initial projects feel too tactical or operational. Significant value your business
can be created by refactoring existing practices. Most importantly, such
early-stage projects provide fertile ground upon which to incrementally
in big and
develop the diverse business, data science, analytics, engineering and small ways.
change management skills required to successfully deploy AI while deliv-
ering incremental value along the way.

Where will you begin?

22
 KIMBERLY NEVALA is the Director of Busi-
ness Strategies for SAS Best Practices. She
is responsible for market analysis, industry
education, emerging best practices and strat-
egies in the areas of advanced analytics, infor-
mation governance and data-driven culture.
Kimberly’s current focus is helping clients
understand the business potential as well as
the practical implications and limitations of
artificial intelligence and machine learning.

A popular speaker and author, Kimberly

About the Author has published numerous e-books and white
papers including: Rationalizing Risk in AI,
Making Sense of AI, The Machine Learning
Primer, Portrait of a CAO, The Anatomy of an
Analytic Enterprise, Sustainable Data Gover-
nance and Top 10 Mistakes to Avoid When
Launching A Data Governance Program.

LEARN MORE
For more insight into AI, visit:
www.sas.com/ai

© 2018 SAS Institute. All rights reserved.

SAS Institute Inc.

100 SAS Campus Drive
Cary, NC 27513-2414
Phone: 919-677-8000
Fax: 919-677-4444

SAS and all other SAS Institute Inc. product or service names are registered trademarks or
trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies.
109891_G86874.0818