Research-Technology Management

ISSN: 0895-6308 (Print) 1930-0166 (Online) Journal homepage: www.tandfonline.com/journals/urtm20

Little ai, Big AI—Good AI, Bad AI

Jim Euchner

To cite this article: Jim Euchner (2019) Little ai, Big AI—Good AI, Bad AI, Research-Technology
Management, 62:3, 10-12, DOI: 10.1080/08956308.2019.1587280
To link to this article: https://doi.org/10.1080/08956308.2019.1587280

Published online: 06 May 2019.

Submit your article to this journal

Article views: 12464

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=urtm20
From the Editor

Little ai, Big AI—Good AI, Bad AI

Jim Euchner

The real risk with AI isn’t malice but strategy or critical process. Creating suc- experience of shopping for clothes.
competence. A super intelligent AI will cessful big AI requires a shift in think- Instead of selecting a specific item, pay-
be extremely good at accomplishing its ing, from a focus on automation or ing for it, and having it shipped to you,
goals, and if those goals aren’t aligned optimization to a focus on reframing you pay StitchFix to predict what you
with ours, we’re in real trouble. strategy based on intelligent capabilities. will want based on your responses to a
—Stephen Hawking The benefits of big AI done well can be questionnaire and your previous expe-
much larger than those of little ai; big rience with the company. The company
Artificial intelligence (AI) is enjoying a AI can—and probably will—transform ships its selections to you, and you
renaissance. Increases in computing entire industries. Advertising, transpor- return any that you do not want. Of
power, advances in machine learning tation, retail, and medical practice are course, this system only works if the
algorithms, dramatic increases in data already being reinvented, with great prediction algorithms are accurate
volumes, and new data structures to benefit to the disruptors. enough to keep the costs of returns
manage the volume are coming together Both little ai and big AI can bring sig- from making the model unprofitable,
to accelerate AI’s application across nificant value to the organization that but the company has completely rein-
industries. The technology has already employs them, but both can also be vented clothes shopping—and it works,
powered improvements in operational implemented badly (Table 1). Little ai is if StitchFix’s performance is any indica-
efficiency, asset utilization, medical usually good when it can be integrated tion. In this case, good big AI is both
diagnoses, and personalized marketing into a work process without fragment- profitable for the company and good for
campaigns, to name just a few areas, ing the work. If, on the other hand, it is the customer.
and its impact is only beginning to be implemented in a way that optimizes a Big AI is bad when its intelligence is
realized. part of the process while sub-optimizing deployed in a way that is detrimental to
Not all AI is created equally, how- the whole, it can result in revenge the customer. A medical insurance
ever. Artificial intelligence can be effects—second-order consequences that application, for example, that seeks to
divided into two categories; I call them can erode or even erase the original predict when medical tests are econom-
“little ai” and “big AI.” Little ai saves benefits (Tenner 1996). ically justified might save money in
money by improving part of an ongoing A system that automates parts of a direct expenses but result in unneces-
operation, perhaps by automating a customer service operation, for exam- sary deaths or larger downstream med-
function currently requiring human ple, can appear to be more efficient by ical costs if it denies tests that would
judgment—performing a diagnostic task a narrow set of metrics, but it can also have detected a condition early.
or optimizing inventory management. be jarring to customers and disempow- Similarly, an AI system that uses its
Big AI, on the other hand, is an enabler ering to employees. The direct cost to massive ability to process personal data
for a complete rethinking of a business the company may be very low, but the is a bad system if it uses that data to
long-term, indirect cost, in terms of cus- manipulate customers rather than to
Jim Euchner is editor-in-chief of Research- tomers who look to other providers out serve them.
Technology Management and Honorary Professor of frustration, could be very high. A The challenge of making AI work,
at Aston University (UK). He previously held senior
better AI system might be designed as whether it’s big or little, is to create
management positions in innovation leadership at
Goodyear Tire and Rubber Company, Pitney an assistant to a human being. That sys- value without losing sight of people.
Bowes, and Bell Atlantic. He holds BS and MS tem would help the human service pro- Given the pace of technology change,
degrees in mechanical and aerospace engineering vider decide what services or remedies achieving this balance will be an ongo-
from Cornell and Princeton Universities,
to offer, without disorienting the cus- ing struggle. Last year, RTM’s parent
respectively, and an MBA from Southern Methodist
University. [email protected] tomer or overriding the human decision organization, the Innovation Research
maker’s power. Interchange, devoted its inaugural
DOI: 10.1080/08956308.2019.1587280
One of the best big AI examples I SPRING futures program to AI and
Copyright © 2019, Innovation Research
Interchange. know of is StitchFix. The company has intelligent systems. Participants and
Published by Taylor & Francis. All rights reserved. used AI to completely redefine the interested parties gathered at a

10 | Research-Technology Management • May—June 2019

TABLE 1. Little ai vs. big AI technologies, he notes, are ready to be
Little ai Big AI deployed today, and exciting new capa-
Good AI Improves prediction or automates a Uses AI to power a business model that bilities are on the horizon. Adams sug-
business task to reduce costs, takes away a customer pain point gests that readers take the time to
speed service, or improve accuracy survey the landscape frequently, seek to
Bad AI Optimizes a function, but makes Implements a new AI-based business understand a wide range of emerging
the whole operation more brittle or model that may improve profits but technologies, and try to understand
less effective; disempowers people manipulates customers connections and synergies among them.
Jaron Lanier, inventor of virtual
conference in Research Triangle Park, intelligent systems but in terms of intel- reality (VR) and tech guru, is the sub-
North Carolina, in October to survey the ligent organizations, of which these sys- ject of this issue’s Conversations inter-
results and consider a range of perspec- tems are a part. Citing some leading view, “What Has Gone Wrong with the
tives on the latest AI renaissance and its examples of companies striving to Internet, and How We Can Fix It.” One
implications. This special issue is drawn become intelligent organizations, he of Lanier’s biggest concerns is the use
from that meeting; the talks adapted notes, “These companies aren’t just ask- of customer data, together with very
here examine the challenges to imple- ing what the technology can do for effective AI, to hyper-target advertising.
menting good AI (or ai) while exploring them; they’re asking what it will mean His belief is that we have moved from
the potential of AI of all sorts to trans- for the technology to be a coworker.” merely targeting consumers to shaping
form business and society alike. Even little ai needs to be designed their behavior, engaging a process
Tom Culver, Lee Green, and Jim together with changes in the organiza- based on the science of behavior mod-
Redden look forward, considering both tion and in collaboration with the peo- ification, one that looks very similar to
utopian and dystopian futures that ple who will work with and alongside suborning addiction. This is a case of big
might result from the evolution of AI. the technology. AI gone bad.
In “Peering into the Future of Intelligent In “Five Lessons for Applying AI is developing very rapidly. As
Systems,” they discuss a process for cre- Machine Learning,” Robbie Allen dis- leaders, we are making critical deci-
ating useful scenarios around an emerg- cusses the results of a survey his com- sions, for our companies and for our
ing technology and also present three pany undertook with more than 150 society, about the proper uses of this
scenarios that can stretch our thinking decision makers in data analytics, seek- technology, often implicitly. Our deci-
about intelligent systems and AI. “Rising ing to assess the state of machine learn- sions will affect how we work, how
Tide,” the optimistic scenario, describes ing. He reviews five lessons about the our companies are organized, and how
a world in which AI, manages to build state of machine learning derived from we interact in the world. Some paths
trust and drives prolonged economic the data, including his perspective on lead to a future where technology sup-
growth. “Stolen Promise” is dystopian, the hype about the topic (it is not over- ports us, providing improved products
describing a future where cyber-attacks blown); the challenges of applying the and services and better workplaces.
and a loss of trust in intelligent systems technology (especially those related to Other paths lead to fragmented work,
leads to information warfare and global data and human resources); and its impoverished social experiences, and
recession. “Moving Target” includes implications for our economy (it won’t a loss of privacy. We hope this issue
both increased innovation and increased cause a jobs problem). He notes that provides innovation leaders with a
insecurity. The choice is essentially one much of the technology is still matur- perspective on both the state of the art
between good AI, bad AI, or a world ing, but he urges companies to move and the stakes of the decisions we
that muddles forward. with urgency to take advantage of near- must make, for this technology will
Brian Bergstein, in “From Intelligent term opportunities. affect all of us.
Systems to Intelligent Organizations,” In his article, “Surfing the Hype
directly addresses the choices we will Cycle, to Infinity and Beyond,” Sam References
make with AI. He discusses positive and Adams discusses the history and future Tenner, E. 1996. Why Things Bite Back:
negative uses of AI technology and pro- of several technologies that are key to Technology and the Revenge of Unintended
poses that we think not in terms of the renaissance of AI. Many of these Consequences. New York: Knopf.

Share your expertise . . .

Contribute to RTM as an author or peer reviewer. Visit us at http://www.iriweb.org/rtm for announcements
of upcoming special issues, author guidelines, and information about volunteering as a peer reviewer.

From the Editor May—June 2019 | 11

 Research-Technology Management seeks submissions
CALL FOR PAPERS

Research-Technology Management is seeking articles that connect theory to practice in the management of innovation.
RTM is actively seeking articles that map the cutting edge in R&D • The common pitfalls in AI efforts
management, illustrate how management theory can be applied • Roadmaps to adoption of AI and automation
in real situations, and give R&D managers the tools to promote • Business model implications of AI and automation
innovation throughout their organizations. We are interested in case
studies, practical models, frameworks, and tools covering a wide Note that we are not interested in technical articles about the
range of issues in innovation and technology manage­ment, particularly technology itself; our focus in this issue will be on how the
open innovation, design-driven innovation, user-centered innovation, technology can be integrated with a wider portfolio as part of a
comprehensive innovation strategy.
and other trends in innovation strategy and R&D management. We
are also seeking articles exploring the future of innovation and how To be considered for the special issue, articles should be submitted by
emerging trends may shape innovation in coming years. June 1, 2019. At submission, select Special Issue Article as the article
type and note the special issue title in the appropriate space.
RTM articles are concise and practice centered. The ideal
submission will provide concrete examples and data to support Special Issue: Open Source Innovation
theories about invention and innovation, the management of
technology and capabilities to support innovation, or the process Open source approaches to innovation and new product
of portfolio selection and management. Successful submissions development are growing in both scope and scale. Open
will offer readers practical information they can put to work source software has assumed a vital role in enterprise software
immediately. development and operations; open source hardware has grown
from niche offerings to a broader phenomenon, spurred by
We prefer submissions at around 4,000–4,500 words, although we the growth of 3D printing; and open source biology is growing
will occasionally publish truly groundbreaking pieces as long as 5,000 rapidly. Open source business models are becoming important
words. Articles should be submitted via our Editorial Manager system at in some industries, and approaches to managing intellectual
http://www.editorialmanager.com/rtm/. For more information, including property are changing as IP commons emerge and growing
author’s guidelines and submission requirements, visit us at www. online communities increasingly support the development and
tandfonline.com/urtm. sharing of IP.
Special Issue: The Artificial Intelligence Renaissance We are looking for articles that describe the opportunities
Artificial intelligence (AI) has emerged from its “AI winter” and and challenges of managing innovation in a world where open
entered a renaissance. Increases in computing power, advances in business models operate alongside closed ones. Specifically,
machine learning algorithms, and dramatic growth in data volumes we are interested in understanding how industrial companies
and data structures to manage the volume have all contributed manage their relationships with open source communities,
to increased success with AI applications and a broadening contribute to the development of open source projects,
of application areas. AI has already led to improvements in compete with or create open source business models, and
operational efficiency, asset utilization, and hyper-personalization in modify IP management practices to accommodate open source
marketing, and its impact is only beginning to be realized. AI and efforts, as well as whether and how they open their products
its related disciplines may well revolutionize innovation, both the to customers and competitors. Of interest are approaches and
products that can be created and the innovation process itself. practices that leaders and teams have used in engaging with
open source approaches and communities, new capabilities that
We are interested in articles about artificial intelligence, machine
are required to enable open source business models, new skills
learning, predictive analytics, cognitive computing, automation,
R&D organizations must develop or acquire to engage with
and related technologies, specifically focusing on how these
open source models, and case studies of companies exploring
technologies can be effectively managed as part of an innovation
open source as a capability.
portfolio. We seek case studies, frameworks, and articles that
provide insight into application spaces, managerial practices, To be considered for the special issue, articles should be submitted
and critical success factors. Topics of interest include, but are not by September 1, 2019. Please select Special Issue Article as the
limited to: article type and note the special issue title in the appropriate
space.
• How to identify applications for AI and related technologies
that make sense for the business Watch this space for more information about forthcoming special
• What is needed for business success with AI issues, or join our email list to receive a notification when calls are
• The implications of AI for business operations released. To join the list, send a request to [email protected].

12 | Research-Technology Management From the Editor