The Future of Business with AI/ML: In Conversation with Tom Davenport, Professor, Writer and Researcher

During an event in New York, Capital One gathered members from the tech community to hear about how AI is changing business across industries. Our guest speaker was Professor Tom Davenport, the President’s Distinguished Professor of Information Technology and Management at Babson College, Co-founder of the International Institute for Analytics, Fellow of the MIT Initiative for the Digital Economy, and Senior Advisor to Deloitte Analytics.

After the event, we sat down with Tom to expand on the conversation and hear his thoughts on where AI is having the most impact -- and has the most implications -- for businesses today. The following conversation has been edited and condensed.

Capital One: What is important to know about quiet, incremental changes companies are making with AI?

Tom: Quiet and incremental changes are the bulk of what companies are doing with AI—including companies like Amazon, where Jeff Bezos said that the majority of their AI projects are “quietly but meaningfully improving core operations.” Each individual project may not be transformational, but if organizations do enough of them they will achieve substantial change.

Capital One: Which specific applications/techniques in AI are you seeing gain the most traction in business realm? 

Tom: Machine learning is the most common technique. It has been around for more than 30 years! In its simplest form machine learning is indistinguishable from predictive analytics. Companies like Capital One have effectively been doing it for a long time. It’s great for predicting customer behavior, identifying fraud, and things like that. 

Capital One: What is one area of AI that’s not widely covered right now, but that you think has the most potential for growth in how businesses will use it in the future?

Tom: Right now there is a need to have a lot of data to train many kinds of machine learning systems. Some companies are beginning to employ algorithms that learn from relatively little data. That could make a big difference in how widely machine learning is deployed. 

Capital One: Do you see it as critical for companies to have their R&D teams? To partner with academics? How important is it for companies to partner with academia to drive R&D? 

Tom: If you want to be on the cutting edge of AI—to be “AI first,” as some companies put it—you definitely need to do a lot of R&D, both on your own and with an external ecosystem. Universities and professors within them are an important part of any R&D ecosystem in AI.

Capital One: Do you see impediments to bridging the gap between breakthrough research and successful business applications?

Tom: The biggest issue is taking a new technology or technique and combining it with an organization’s existing systems and processes. That integration is something that many companies say is their greatest barrier to AI innovation. Accumulating the necessary data to train your models is often a problem too.

Capital One: Broadly speaking, what are some of the major upsides for the average consumer/customer when you think about the benefits of AI from a customer standpoint?

Tom: There are a variety of smaller changes in personalization, fraud detection and so on that consumers will benefit from. But they won’t necessarily notice a lot of the small improvements. But for the big changes—the revolutionary ones--they need to be patient! The transformational applications of AI take a long time to come about. Think about autonomous vehicles, for example. We have been talking about them—and expecting that they are just around the corner—since the mid-1980s. They will eventually get here, but a lot has to change first—not only technology, but regulation, infrastructure, etc.

Capital One: What is one key risk that think companies need to be aware of in terms of their AI strategy?

Tom: It’s pretty important for companies to have patience too, and not expect too much too soon. They can and should have ambitious goals for AI, but they should realize that it will take a long time and many projects to realize them. The greatest tragedy for enterprise AI would be if everyone became disillusioned about how hard it is or how long it takes to have a substantial impact. I don’t think we will have another “AI winter,” but it’s important to be realistic about what the technology can deliver. Over time it will be revolutionary, but in the short run it’s evolutionary.

Capital One: What role do you see considerations like transparency, accountability, and responsibility play when it comes to corporate responsibility of AI?

Tom: They are all critical, and exploring them is just as important as increasing a technique’s predictive power. Customers won’t want to deal with companies that can’t give them an explanation for key decisions and actions, and they want to be sure that there is no unfair bias in algorithms. And regulators care very much about this too. So I think every company that makes substantial use of AI should have a framework for ethics and a governance approach for monitoring it and setting policy. 

Capital One: In your view, how do considerations around responsible AI impact the hunt for talent in this space? Should firms be complementing the technical experts with legal, social science, design, and other related experts when thinking of technology development?

Tom: Absolutely. You have to have the analytical and technical skills, but also other types of domain knowledge. With new tools like automated machine learning, much of the really technical work can be done by machines. That leaves humans to have the domain knowledge, the change management skills, and the ability to integrate AI with the current process and workforce.

Capital One: What strategies do you recommend to large companies as they navigate the build vs. buy conversation in terms of AI strategy?

Tom: It’s no different than other technologies. If you want a really distinctive capability, you have to build it. If you just want some basic functionality for something like lead scoring for salespeople, it’s usually smarter to buy it from a vendor. That way you already have the necessary data, and your users are familiar with the interface. Most companies should try to identify where they need real differentiation, and focus their “build” efforts there. They can buy everywhere else if there is an available vendor that does good work.

Capital One Tech
Capital One Tech,

Stories and ideas on development from the people who build it at Capital One.

DISCLOSURE STATEMENT: © 2019 Capital One. Opinions are those of the individual author. Unless noted otherwise in this post, Capital One is not affiliated with, nor endorsed by, any of the companies mentioned. All trademarks and other intellectual property used or displayed are property of their respective owners.

Related Content

Machine Learning

Start Your Engines: Fueling Escalating Technologies in Machine Learning

Machine Learning

Using Machine Learning to Re-Think the Customer Experience Paradigm

Machine Learning

Machine Learning in the Enterprise: Lessons from the Front Lines