Académique Documents
Professionnel Documents
Culture Documents
The Road to AI
August 2018
Digital Business
EXECUTIVE SUMMARY
Artificial intelligence is slowly but surely becoming a fundamental capability for organizations.
Businesses are learning to use new AI technologies – everything from chatbots to neural net-
works – to create personalized experiences, intelligent products and smarter business processes.
But the path to AI is long and winding. Every application requires different tools and algorithms,
and unlike other scalable technologies such as cloud and analytics, AI requires a fresh look at
every use case.
To ensure AI efforts achieve their desired business outcomes, we recommend the “5 E” process:
• Educate up and down the organization. Fulfilling AI’s potential means knowing how to apply it
to the right business problems and processes.
• Embrace experimentation. With AI, openness to new business needs is as important as trying
out new technologies and techniques. Yet AI experimentation differs from other forms of digital
in several significant ways.
• Evaluate. Are the AI pilot results definitive? What’s the next phase for AI within the organization?
Because nearly every organization is still learning how to apply AI, it’s easy to get lost in the
evaluation process.
• Establish priorities. Given AI’s growing profile, determining the projects that offer real business
value is a key step for organizations.
• Explore further. Building out AI as a capability means tackling thorny questions such as whether
to focus AI efforts within a single functional area or apply them broadly across the company.
There is no utility AI. Every company must chart its own path to success. This white paper offers
insights and recommendations to help organizations identify use cases, pilot and learn from
them, and then repeat the process so AI becomes a key element in the organization’s continuous
improvement.
2 | The Road to AI
Digital Business
The Road to AI | 3
Digital Business
Getting up to speed on AI involves moving iteratively along a maturity curve, and for good reason:
Every application and use case requires different tools and algorithms, and every organization is at a
different place in terms of its AI maturity curve. The chatbot for a financial service provider’s IT help
desk, for example, can’t be applied to the same company’s home-loan call center. An algorithm trained
in pharma to read documents for adverse event recognition can’t be reused in a banking context to
identify anomalies in mortgage applications, even though the technology stack and technique may be
the same. Choosing the right next-level AI applications propels an organization forward on its AI
maturity curve.
With AI, each use case requires a unique training process as the system learns the relevant patterns.
Moreover, AI systems take much longer to master some tasks than others. For example, neural net-
work-powered computer vision requires extensive training and data sets to recognize and analyze
patterns in images.
AI also offers a very different experience from other digital exploits. While businesses can scale their
learning of cloud and analytics, AI requires a fresh look at existing approaches to help take advantage
of new techniques, different data sets and accelerating advances in core technologies.
Many organizations we work with have already embraced AI elements and are achieving meaningful
business outcomes:
• At a financial services company, fraud detection is 25% more accurate as a result of the real-time
algorithms our teams created.
• An energy company is saving $1 million annually per oil rig due to the predictive maintenance
program we developed.
• A leading insurer has netted a $30 million topline increase and 20% lift in lead conversion when
it launched a new customer activity hub.
(For more detail on these engagements, as well as how we’re helping our clients apply AI in their busi-
nesses, please visit the “Featured Work” section of the Cognizant Digital Business, AI & Analytics
section of our website.)
4 | The Road to AI
Digital Business
Getting up to speed on AI
involves moving iteratively
along a maturity curve, and for
good reason: Every application
and use case requires different
tools and algorithms.
The Road to AI | 5
Digital Business
1 Educate
With AI on every organization’s 2018 to-do list, companies are attending conferences and hosting
workshops. Despite the buzz, confusion remains over even the basics of AI. AI requires its own literacy.
For example, how does an advanced form of AI such as machine learning (ML) differ from other forms
of the technology? (See Quick Take, below.)
In addition to these distinctions, it’s important to understand that AI forms a continuum: There is no
start or end, and it’s the combination of tools and techniques, applied to the right business problems
and processes, that will deliver personalized experiences with efficiency and scale.
QUICK TAKE
What AI is:
• Making intelligent, human-like decisions based on facts.
• Pairing decisions with actions that either automate a human task or improve
experience.
• Interpreting unique human traits such as natural languages, speech patterns and
images.
• Mimicking human intelligence by analyzing and acting on structured data that’s
machine- and application-generated.
6 | The Road to AI
Digital Business
2 Experiment
To succeed with AI, businesses need to be open to discovering new and, at times, unexpected business
needs. For example, when a healthcare client recently conducted an AI pilot in natural language pro-
cessing (NLP) to more efficiently review social workers’ and physicians’ notes, the company was
willing to probe for nuances in health outcomes that no one was yet looking for. The organization
discovered that the health of 12,000 patients in a pilot market was significantly affected by factors
such as economics and access to transportation – yet standard documentation practices included no
fields for such social factors. The finding is important to the quality of the company’s patient care
because non-health factors impact 40% of wellness outcomes.2 Without the healthcare provider’s
openness to new ideas, it would not have uncovered such important information.
In AI, experimentation begins with the willingness to view data holistically. What’s the underlying root
cause for the findings? How can the attributes be tracked in the system? Businesses need to rethink
how work is done, identify the new business structures needed to support this work, and spot the
resulting opportunities to grow revenue and improve performance. All of this requires the willingness
to trust machine intelligence – no small feat within many business functions.
To their credit, many organizations have grown comfortable experimenting with digital tools and tech-
nologies and look forward to doing the same with AI. But AI differs in several areas.
• ROI-driven outcomes. How will an AI experiment apply to the business? More often than not,
organizations’ AI efforts emphasize technology capabilities and algorithms rather than business
impact and benefits. Measuring business value from the get-go is critical to prove an experiment’s
ROI – even if it means the experiment is deemed a failure. For example, an AI pilot that is successful
from a technological standpoint may deliver business outcomes that are too meager to justify the
process and cultural changes required.
• Setup and support. AI requires a fail-fast, learn-faster environment. Organizations need processes
and ecosystems to support AI experimentation. For example, many companies are unprepared for
AI’s abbreviated development window. Experiments can be up and running in four weeks, and
pilots can be rolled out in four to eight weeks. Yet we recently worked with a company that spent
five months conducting a machine-learning pilot because it lacked the infrastructure and pro-
cesses to support a quick turnaround.
How can organizations better prepare for rapid AI experimentation? Ready access to new technol-
ogies and techniques is an important first step. Many AI efforts get bogged down in lengthy
technology procurement times. Businesses need to be sure they have an open cloud environment
to experiment with machine data. Better yet, they should create a robust set of partnerships that
provide access to continuously advancing AI technologies. Because no one organization can do it
alone when it comes to AI, businesses should examine the good work happening in start-ups and
with other large providers and consider where collaboration can strengthen their AI efforts.
The Road to AI | 7
Digital Business
Indeed, there is so much to learn in AI that every experiment is a stepping stone. A good example
is the work done by one of our energy and utility clients to reduce customer attrition with a
machine-learning system. We developed an outcome-oriented experiment that investigates new
algorithm techniques and probes data for more details on why consumers switch providers. Apply-
ing an ethnographic approach helped us to understand how and why consumers make energy
decisions. The approach points to new, untapped data sets for further investigation. Is the project
a success? Possibly. Together with the client, we defined the experiment’s success as improved
algorithmic predictability and a better understanding of AI techniques and the data’s value. The
project is ongoing, and while it might not result in the explicit answers our client is seeking, it will
deepen the AI expertise and understanding of the business problem for all involved by viewing the
issue through a human lens.
3 Evaluate
It’s easy for organizations to get lost here. Determining whether a pilot has produced definitive results
is tricky, as is the question of whether to extend a pilot for further iteration or to acquire additional
data sets. For example, a client that provides credit card services to small and medium businesses
(SMBs) discovered that while it typically segments customers by industry and revenue, a more telling
metric is whether the SMB’s founder is still involved. When the original owners remain hands-on, the
SMB often has little time to evaluate new financial products. The client’s next step is to determine
whether the additional campaign’s ROI will offset the costs. The lesson? Be willing to stop the pilot if
the cost-benefit doesn’t work out.
8 | The Road to AI
Digital Business
Even thornier questions arise after the initial pilot assessment. Within the organization, what’s the
next phase so that AI adds value? Once a company has engineered a successful AI pilot, how can it
propagate that experience through the organization?
At many businesses, the evaluation stage can be a tug of war: IT wants to push AI initiatives forward to
demonstrate its proficiency with leading-edge technologies, while the business prefers to wait to under-
stand the implementation more fully. Promoting more transparent testing of technologies and techniques
can break the gridlock, enabling organizational leaders to observe project successes as they occur and
helping them to feel more comfortable about moving projects from the lab into the business.
Organizational constructs to oversee AI initiatives are important to achieve the necessary consen-
sus. Some companies establish separate teams within the innovation function, while others form
joint AI councils across IT and business units. Because data is key for AI, some businesses add the
AI mandate to their data organization. The idea is not to add to the org chart but to better under-
stand how everyone in your organization can learn from each other and avoid repeating mistakes.
The key is to create a nimble organization in which all stakeholders – business owner, process owner,
data owner and technology owner – come together to experiment with business outcome- focused
use cases (see Figure 1).
Establishing
an AI office
The Road to AI | 9
Digital Business
Regardless of where AI is housed in the org chart, most companies recognize the need for multifunc-
tional participation. AI isn’t an island, and it can’t be a skunkworks effort. It requires a specific business
need, and unlike the nascent stage of other technologies – think blockchain or quantum computing –
AI has followed a speedy trajectory from cool technology to prospective business solution. Thanks to
its appearances in well-known demonstrations – such as Google DeepMind’s triumph in the board
game Go4 and IBM Watson’s 2011 win on TV’s Jeopardy5 – AI has fired up our collective imaginations.
Yet organizations still struggle with the question of which business use cases are best for AI, and how
to know whether they’re working.
4 Establish Priorities
Given AI’s growing profile, it’s common to find multiple business units within an organization – opera-
tions, technology and lines of business – each pursuing its own initiatives. Use cases abound, and
prioritization is a challenge. Which pilots share a common AI core that all functions can leverage?
Which ones can the larger company learn from? The end goal is to establish AI as a capability that the
organization as a whole can embrace.
To encourage AI experimentation while imposing order and discipline on the prioritization process,
CIOs and business leaders can ask several questions: Does a proposed project deliver limited, incre-
mental value, or is it reinventing a process through clever use of data and technology? Early successes
that feature positive business benefits, such as a boost to the top or bottom line or productivity
improvements, help organizations embrace AI faster than those that are technologically possible but
have limited value. (See Quick Take, page 11.)
After starting with the question of business value, businesses should then move on to technical feasi-
bility. Does the prospective AI system have the data it needs from which to learn patterns? Is the data
free of bias? Is the technology infrastructure able to process different types and large volumes of
data? If the answer to any of these questions is no, then the project’s technical feasibility is a limiting
factor. (For more information on generating business value from AI, read our white paper “AI: Ready
for Business.”)
5 Explore Further
In the exploration phase, organizations are typically deciding whether to continue focusing their AI
efforts within a single functional area, or to apply them more broadly across the company. Many are
also watchful of unfolding government regulations regarding compliance and liability, as legislative
and judicial branches tackle AI-related challenges. The important part of this step is for companies to
consider how they can better organize themselves around AI. What processes can they create that are
useful for applying their AI learnings to other parts of the organization? (See Quick Take, page 12.)
10 | The Road to AI
Digital Business
QUICK TAKE
Vision to Value
When a biotech company launched a pilot that applied machine learning to
natural language processing, it wanted to explore the technology’s feasi-
bility. More important than the AI techniques, however, was the company’s
vision for the pilot to contribute to its mission of improving health outcomes.
It’s the job of the company’s patient services group to stay in close tele-
phone contact with individuals who have been prescribed its specialty drugs.
To better understand the drugs’ efficacy, case managers speak regularly
with patients and document their experiences. The teams take notes on
each conversation. The metrics are similar to a call center.
The pilot’s results have helped the company double-down on its mission.
It led to new efficiency measures, such as more automated note-taking. It
also helped zero in on patients at risk of noncompliance with drug regimens
and opened opportunities for proactive intervention. Equally important, the
results prompted the company to expand its case-manager training to include
greater emphasis on empathy for patient concerns. The new approach to
training has the dual benefit of potentially improved health outcomes for
patients and greater job satisfaction among case managers.
The Road to AI | 11
Digital Business
QUICK TAKE
• Small, multi-skilled teams are critical. AI success depends on combining knowledge from
business functions, processes, data and technology. It takes an organizational village.
• Speed is of the essence. AI’s rhythm is to pilot, learn and scale. To make it happen, you need
to assemble the relevant skills and teams to work quickly and iteratively.
• Closing the learning loop is essential. Because learning happens on multiple fronts, it
unlocks new capabilities and approaches that can be applied to other parts of the business.
It’s important to have an organizational construct that can oversee multiple AI experiments
in parallel and still ladder up to a centralized approach to learning that advances core busi-
ness capabilities.
• Never forget humans are at the center of all key business initiatives. Balancing human
ambition with machine resilience enables AI to grow. It can’t be an afterthought. Focus on
finding balance from the get-go by emphasizing continuous AI literacy, skill retraining and
role retooling.
12 | The Road to AI
Digital Business
LOOKING AHEAD
When it comes to digital pursuits, there’s nothing like AI. Rather than applying their learnings from
other digital initiatives, businesses need to get ready for a whole new way of thinking to reap the full
success that AI can offer. From rethinking old ways of work, to recognizing new types of value,
AI requires a fresh look at existing approaches.
Businesses can develop the mindset that will instill success by educating their workforce on AI,
embracing experimentation, understanding how to evaluate AI pilots, determining project prioritiza-
tion, and pushing AI insights further into the organization. While each business will take its own path
to AI, all organizations can follow this process to optimize business results.
To learn more, please visit the AI & Analytics section of our website.
The Road to AI | 13
Digital Business Digital Business
FOOTNOTES
1 Henry Stewart, “Eight Companies that Celebrate Mistakes,” Happy, June 8, 2015, https://www.happy.co.uk/8-compa-
nies-that-celebrate-mistakes/.
2 Steven A. Schroeder, “We Can Do Better — Improving the Health of the American People,” The New England Journal of Medi-
cine, Sept. 20, 2007, https://www.nejm.org/doi/full/10.1056/nejmsa073350.
3 Henry Stewart, “Eight Companies that Celebrate Mistakes,” LinkedIn, June 8, 2015, https://www.linkedin.com/pulse/8-compa-
nies-celebrate-mistakes-henry-stewart/.
4 Jon Russell, “Google’s AlphaGo AI Wins Three-Match Series Against the World’s Best Go Player,” Techcrunch, May 25, 2017,
https://techcrunch.com/2017/05/24/alphago-beats-planets-best-human-go-player-ke-jie/.
5 Kate Torgovnick May, “How Did Supercomputer Watson Beat Jeopardy Champion Ken Jennings? Experts Discuss,” TED Blog,
April 5, 2013, https://blog.ted.com/how-did-supercomputer-watson-beat-jeopardy-champion-ken-jennings-experts-discuss/.
14 | The Road to AI
Digital Business
The Road to AI | 15
ABOUT COGNIZANT DIGITAL BUSINESS
Cognizant Digital Business helps our clients imagine and build the Digital Economy. We do this by bringing together human insight, digital
strategy, industry knowledge, design, and new technologies to create new experiences and launch new business models. For more informa-
tion, please visit www.cognizant.com/digital or join the conversation on LinkedIn.
ABOUT COGNIZANT
Cognizant (Nasdaq-100: CTSH) is one of the world’s leading professional services companies, transforming clients’ business, operating and
technology models for the digital era. Our unique industry-based, consultative approach helps clients envision, build and run more innova-
tive and efficient businesses. Headquartered in the U.S., Cognizant is ranked 205 on the Fortune 500 and is consistently listed among the
most admired companies in the world. Learn how Cognizant helps clients lead with digital at www.cognizant.com or follow us @Cognizant.
© Copyright 2018, Cognizant. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted in any form or by any means,electronic,
mechanical, photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein is subject to change without notice.
All other trademarks mentioned herein are the property of their respective owners.
Codex 3614