Interview with Thomas H. Davenport
ChatGPT burst upon the AI landscape in November 2022 with a media and market frenzy not seen since Steve Jobs introduced the iPhone in 2007. Over one million people immediately signed up to test if OpenAI’s large language learning algorithm could perform as well or better than a human at answering text-based research questions, writing a poem in the style of Shakespeare or songs like Bob Dylan, completing high school homework assignments (you can imagine the handwringing here), and much more. In many cases, the answer was yes.
Only three months later, there have already been three major upgrades to the search and auto correct algorithm. ChatGPT4 gets the answers wrong and “hallucinates” (makes up stuff) less frequently. Competitors from Google and many others have joined the race to use AI to empower everything we do online and virtually every process we use inside organizations.
As we all know, the promise of AI won’t be actualized unless we develop a strategy, explore use cases, and operationalize the technology in our organizations. This takes organizational savvy and change management skills, which no algorithm can give you.
Fortunately, we can learn from multifaceted experts like Tom Davenport and a decade of early adopters who have used AI to dramatically accelerate their businesses. In addition to dozens of books and articles on topics like knowledge management and analytics, Tom is co-author (with Nitin Mittal) of All in On AI: How Smart Companies Win Big with Artificial Intelligence. In this interview excerpt, Tom speaks with me about what it means to be “All in” on AI, shares some key lessons from early adopters, and provides insight into what AI will mean for the workplaces of the future.
What does it mean to be “all in” on AI?
Organizations that are “all in” on AI are highly invested in a variety of different ways. One is that they have different types of AI spread throughout the company quite broadly. We’re talking about dozens of use cases at a minimum, but more commonly it’s hundreds or even thousands of use cases because AI is a narrow technology. It tends to support or automate tasks, not entire jobs—and certainly not entire processes. So for example, if you want to automate all customer service or all order management, you will need to assemble many different pieces of AI to have a high level of impact.
Being all in means you’re not just using AI for optimization at the margins of your business but really changing something dramatically. In the book, we talk in terms of organizations that use AI to transform their strategy, business model, products, services, operations, or even customer behaviors—not just making small tweaks for operational improvement.
You’re probably going to get into trouble unless you’re ethical about the ways in which you use AI. Organizations that are all-in also have a framework for ethical and trustworthy AI in place that includes guidelines, policy approaches, and governance structures.
What led you to focus on early adopters of AI in your book?
I wrote a book called Competing on Analytics in 2007. My readers found it quite helpful to look at companies that were aggressive early adopters of analytics, and I thought that the same thinking probably applies with regard to AI. The people I was working with at Deloitte were starting to talk about the ways in which AI could help transform these big legacy companies, a lot of whom were their clients. I think the nice thing about those examples is that even if people can’t or don’t want to go all in, they respond to reading about what it’s like to be all in and what companies can accomplish if they’re really aggressive in their adoption of the technology.
What lessons can we learn from these early adopters?
The easiest and most concise piece of advice I always give is to think big, but start small. There are all these little pieces to AI, but you have to think about it in a strategic way. You want to accomplish something dramatic with it, not just improve at the edges of your business.
We can make an analogy to the early days of electricity. Some textile mills and other factories used electricity by having a big drive shaft running down the middle of the building, and big belts attached to all the machines that needed power. It took around 20 years before they figured out that they could just plug electricity into each of those small machines. Let’s hope we’re not making the same mistakes with AI.
Another important lesson is to understand your process and try to improve it before implementing AI or related technologies. There are a lot of companies that have created centers of excellence for automation and AI, and most of them also involve process improvement ahead of time. They all seem to realize that it’s stupid to automate a bad process. Figure out if you can eliminate steps or improve the process first.
Where do humans fit within an AI-fueled organization?
AI is already starting to change what we think about as uniquely human capabilities. But we haven’t yet seen any examples of AI turning against us. If that happens, I think all bets are off. But right now I think it’s much more about augmentation than it is about eliminating humans from the picture.
We’re changing the way people do their work, but that does not eliminate the need for humans. Certainly I think that’s true of AI’s generative capabilities. If you are silly enough to think that you can generate a school paper or a blog post or anything else without having to read and edit it, you’re not going to keep your job for very long. So I think there is still plenty of room for humans.
How can organizations get ready for AI?
Organizations should tell their employees that their job is going to be changed by AI and that it’s important to get ready for that change. The message needs to be that we’re confident that if you keep up and figure out how to use the technology to add value, you have a job here for as long as you want one. But if you don’t, you may not have a job.
Organizations also need to encourage people to experiment. As I looked at case studies of people working with AI, most of them are heat-seeking missile-types who like to try out new things and use all the latest technologies in their personal lives. I think everybody needs to do that. It’s not easy—it takes time and effort—but I think it’s worth the trouble, and that’s the world we live in, like it or not.
Read Carla O’Dell’s full interview with Tom Davenport here. For more insights from Tom Davenport, see: Smart Machines: Could This End Badly for Knowledge Workers? (2015 interview with Carla O’Dell).
Interview with Thomas H. Davenport