Digital Business

The Road to AI Incorporating artificial intelligence into your business systems and processes is a journey unlike any other digital technology implementation. Here is a five-step process for navigating it successfully.

July 2018

2 / The Road to AI

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 networks – 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.

The Road to AI / 3

Digital Business

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.

4 / The Road to AI

Digital Business

The path to AI success

There’s no such thing as general-purpose AI. Every company’s artificial intelligence journey is unique, as is its application of AI. Maybe the business has invested in big data systems and wants to apply AI but is unsure where to start. Maybe it’s at the stage of identifying the areas in which AI can add maximum business value. Perhaps it’s finalizing the underlying technology stack.

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 network-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 businesses, please visit the “Featured Work” section of the Cognizant Digital Business, AI & Analytics section of our website.)

5

The Road to AI / 5

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.

6 / The Road to AI

Digital Business

Achieving outcomes with the 5 E approach

With the 5 E approach, businesses can ensure their AI efforts are as streamlined and successful as possible, and that they achieve the desired business outcomes.

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.

1

Quick Take

What AI Is – and Isn’tFirst, a word about what AI is not. It’s not:

• Using machine learning to find correlations and patterns in data.• Deploying machine learning to generate predictions captured in a report.• Implementing cognitive technologies to extract information from text, speech or

images, with no associated action. • Applying pre-defined rules to automate human tasks.

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.

The Road to AI / 7

Digital Business

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 processing (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.1 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 technologies 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 processes to support a quick turnaround. How can organizations better prepare for rapid AI experimentation? Ready access to new technologies 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.

2

❙ Increased tolerance for failure. The spirit of experimentation embraces the idea that not all initiatives will pan out. With AI, every organization will make some bad bets. Not only is acceptance of failure key, but failure in AI/ML isn’t binary: Sometimes pilots are technological successes but yield few benefits. Perhaps AI will inspire more organizations to view failure as a badge of honor. When Google announced it would shut down its online platform Wave just a year after its launch, it rewarded the development team for having taken a calculated risk. Accounting software company Intuit holds “failure parties.”2 “Every failure teaches something important that can be the seed for the next great idea,” says co-founder Scott Cook.3 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. Applying 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.

8 / The Road to AI

Digital Business

With AI, every organization will make some bad bets. Not only is acceptance of failure key, but failure in AI/ML isn’t binary: Sometimes pilots are technological successes but yield few benefits.

The Road to AI / 9

Digital Business

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.

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 understand 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 consensus. 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 understand 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).

How should we structure the AI organizationto address the business’s LOB requirements?

What are the key roles necessaryto set up and operationalize projects within the AI o�ce?

What are the frameworksavailable for PoVs (proofs of value) vs. pilots as well as projects?

What are the top guiding principlesthe AI o�ce should establish?

How should we approachchange management and

business stakeholdermanagement?

How should AI solutionsbe governed for

accountability & reusability?

Establishingan AI office

Organizational considerations for establishing an AI office

Key questions to ask when planning your next AI initiative.

Figure 1

Digital Business

10 / The Road to AI

Regardless of where AI is housed in the org chart, most companies recognize the need for multifunctional 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.

Establish priorities

Given AI’s growing profile, it’s common to find multiple business units within an organization – operations, 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, incremental 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 feasibility. 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.”)

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.)

4

5

Vision to ValueWhen a biotech company launched a pilot that applied machine learning to nat-

ural language processing, it wanted to explore the technology’s feasibility. 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 telephone con-

tact 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 conver-

sation. 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 biotech company’s use of AI is a helpful example of vision to value:

Examine the organization’s strategic value and how the AI pilot connects to it.

The company is now using the successful results to cultivate AI literacy across the

organization. It’s showcasing the value of new techniques to transform patient

engagement as well as to create “aha” moments in business leaders’ minds about

the possibilities of AI/ML.

Quick Take

The Road to AI / 11

Digital Business

Quick Take

AI’s Recombinant DNAsPerhaps even more than the right technology, AI requires the optimal blend of

business case and corporate culture to succeed. In many ways, it reshapes the

companies that adopt it. The following ingredients are essential to creating an

effective AI culture:

❘❘ 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 business 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.

❘❘ Communicate, communicate, communicate. Sharing knowledge and experience

is at the heart of corporate AI efforts. Apply it to success and failure. Use constructive

words that convey and reinforce business value, benefits and outcomes through the

use of technology, superior techniques and differentiated data.

12 / The Road to AI

Digital Business

The Road to AI / 13

Digital Business

Looking aheadWhen 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 prioritization, 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.

Digital Business

14 / The Road to AI

Endnotes1 Steven A. Schroeder, “We Can Do Better — Improving the Health of the American People,” The New England Journal of

Medicine, Sept. 20, 2007, https://www.nejm.org/doi/full/10.1056/nejmsa073350.

2 Henry Stewart, “Eight Companies that Celebrate Mistakes,” Happy, June 8, 2015,, https://www.happy.co.uk/8-companies-that-celebrate-mistakes/.

3 Ibid.

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/.

The Road to AI / 15

Digital Business

About the author

Poornima RamaswamyVice President, Cognizant Digital Business’s AI and Analytics Practice

Poornima Ramaswamy is Vice-President of Cognizant Digital Business’s AI and Analytics Practice. With her 20 years of experience, she consults and works with clients across industries in North America to leverage their vast amounts of data and convert it into meaningful insights to improve revenue goals and drive business process efficiencies. Her focus has been to help clients in their analytics and AI transformation journey and help them transition from a data-driven business to an insights-driven business.

Poornima also runs Cognizant’s Chief Data & AI Officer Advisory Council, which is a community of analytics executives who focus on making AI/analytics a strategic imperative in their organization. She has an MBA (technology and finance) and an undergraduate degree in mathematics. Poornima can be reached at Poornima.Ramaswamy@cognizant.com | www.linkedin.com/in/poornima-ramaswamy-4b97021/.

About the author

© 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.

TL Codex 3614.02

World Headquarters

500 Frank W. Burr Blvd.Teaneck, NJ 07666 USAPhone: +1 201 801 0233Fax: +1 201 801 0243Toll Free: +1 888 937 3277

European Headquarters

1 Kingdom Street Paddington Central London W2 6BD EnglandPhone: +44 (0) 20 7297 7600 Fax: +44 (0) 20 7121 0102

India Operations Headquarters

#5/535 Old Mahabalipuram RoadOkkiyam Pettai, ThoraipakkamChennai, 600 096 IndiaPhone: +91 (0) 44 4209 6000Fax: +91 (0) 44 4209 6060

About Cognizant Digital BusinessCognizant 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 information, please visit www.cognizant.com/digital or join the conversation on LinkedIn.

About Cognizant Ai & AnalyticsAs part of Cognizant Digital Business, Cognizant’s AI & Analytics provides advanced data collection and management expertise, as well as artificial intelligence and analytics capabilities that help clients create highly-personalized digital experiences, products and services at every touchpoint of the customer journey. We apply conversational AI and decision support solutions built on machine learning, deep learning and advanced analytics tech-niques to help our clients optimize their business/IT strategy, identify new growth areas and outperform the competition. Our offerings include AI to Insight, Customer Intelligence, Intelligent Automation, Product Intelligence, and Risk & Fraud Detection. To learn more, visit us at www.cognizant.com/cognizant-digital-business/applied-ai-analytics.

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 innovative and efficient business-es. Headquartered in the U.S., Cognizant is ranked 195 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.