AI + Leadership
We Will See AI-Native Companies Emerge, Just Like The Digital-Native Firms Of The Past
AI + Leadership
Jon Iwata, Executive Chair of the Data & Trusted AI Alliance, discusses the real risks of inefficiency, guides versus guarantees, and reinvention with David Reimer and Adam Bryant, in the latest addition to the AI + Leadership series.
Reimer: Can you share a quick overview of how the Alliance was formed, and how it’s evolved?
Iwata: I spent 35 years at IBM, which has been an important player in many technology cycles over its more than 100-year history. When I joined the company, the personal computer was the big conversation. I was there when we sold off the PC business. I lived through the internet era and cloud computing and the earliest days of AI.
In 2019, Ken Chenault, the former CEO of American Express who had been a longtime director at IBM, was also on the board of Facebook, which was dealing with its controversy involving Cambridge Analytica. Ken is a very principled guy, and he recognized that this was not a PR or government relations issue. They were fundamental business practice issues. That led to a series of discussions about how businesses will use data and AI and, ultimately, the formation of the Data and Trusted AI Alliance, which Ken co-chairs to this day with Sam Palmisano, the former CEO of IBM.
The Alliance is made up of mostly large companies across a number of industries, and our goal is to accelerate learning about the responsible use of data and AI. Over time, our focus has shifted to how businesses can extract value from their investments in AI.
Bryant: On that point, there seems to be two areas of focus for extracting business value. One is to take costs out through greater efficiency, and the other is to create greater value through innovation.
Iwata: Historical patterns are helpful here. The first use of a new technology tends to be productivity. We saw that with the PC. And when the internet moved into wide use, it dramatically cut the costs of transactions. It was an irresistible productivity tool because you could make everything essentially self-service.
I found a data point from the 1990s that the cost of a transaction in a physical bank branch was more than $2. The ATM lowered that cost to about 25 cents. And online, it was only fractions of a penny. Companies across all sorts of industries, including retail and airlines, saw similar savings. And employees were also able to adopt a self-service approach inside companies. The internet was a massive productivity tool.
But then in the 2000s, we saw companies spring up like Airbnb, Uber, Spotify, Netflix, and PayPal. These were not tech companies in the traditional sense. They took advantage of the available digital capabilities, created new kinds of business models, and disrupted the incumbents. The market values of those companies compared with the incumbents is staggering.
Now, let’s talk about the bubble. Broadly speaking, there were two sets of peers. One was the tech providers. Remember that Cisco was the most valuable company in the world during the internet build-out. Cisco and Microsoft survived, of course, but Netscape and Sun Microsystems didn’t.
But the dot-com bubble really wasn’t about Cisco, Microsoft, and Sun Microsystems. It was about the dot-com companies with unsustainable business models. Now let’s take those lessons from history and apply them to today. People are questioning the soaring valuations of companies like Nvidia and OpenAI. We’ll see if there will be a correction there.
But the real question is, what are most businesses going to do with AI? Right now, just as we saw with the internet, it’s primarily about productivity. But it’s inevitable that we are going to see the equivalent of the digital-native companies of the past emerge as AI natives—AI-native banks, AI-native pharma, AI-native retail, and so on. So, when you talk about getting value from applied AI, it’s going to come from both productivity and business innovation. Clearly, the latter is nascent.
Reimer: One of the things that can get lost in the breadth of possibilities is specificity—what do we do next? You’re not talking about growth in the abstract but thinking quite concretely about how we might actually use AI to blow up our own models.
Iwata: People sometimes think that the opposite of productivity gains is revenue growth. But that’s not what happened with Netflix and Spotify, Airbnb, and Uber. They didn’t just take away market share from others. They redefined value chains and they redefined markets.
If you can use AI to boost sales and grow market share from your defined competitors within your defined market, great. But the bigger threat is that you will no longer have that market because someone else is going to come in and redefine it.
Bryant: How have you seen organizations embrace AI for creativity and innovation, not just doing their jobs faster?
Iwata: All the companies in the Alliance are using AI for software development. We have a peer group of chief software officers, and they compare notes. They are seeing, in some cases, impressive productivity gains. However, they are also seeing the limits of that.
Again, this is a familiar pattern when you optimize existing processes and workflows. You hit a limit because what’s needed is a complete redesign of the workflow. People are now recognizing that they have to rethink the entire workflow of the software lifecycle to really get value from AI.
Some companies assume that a successful pilot can be scaled, which is a false assumption because you need to think about which way you’re going to scale it. Scale it up? Scale it out? Those discussions lead to having to redesign the work and rethink roles. That’s just a much bigger lift, and not everyone is up for it.
Reimer: What are the evolving implications for leaders, particularly those who are trying to build the right workforce for the future?
Iwata: Again, history is helpful here. Some job categories will disappear entirely, new ones will be created, and everything in between represents reskilling. How it actually happens begins with new tasks, and then the new tasks sometimes aggregate into roles. Then the roles may become standardized into new jobs, and they may even become professions.
I’ll give you an example. In 1995, at IBM we had to figure out this new www.ibm.com thing. We had to register URLs all over the world. Then we realized we had to have a website. Then we had to have content for the website, and people to manage it. Task after task. That eventually created the role of webmaster. I believe we’ll see the same progression happen with AI—new tasks are beginning to aggregate into new roles.
Bryant: Was there are a moment when you personally were surprised by what AI can do?
Iwata: I use five of these tools routinely, some for personal use and some for work. I usually have multiple chats and projects going at any given time. There are moments when it will make a recommendation or give me an answer that takes into account something I asked it to do nine months earlier. It’s a bit disturbing but I’m also amazed. Its deep knowledge of us as individuals is remarkable.
Reimer: What is it in your background that has made you comfortable in, to use a metaphor, waters that are not all flowing in one direction?
Iwata: My parents were born in 1932 in Hawaii, and they never complained about their childhood. They thought it was magical, and in many respects, it was. But they grew up in the Great Depression and then lived through World War II. They were skeptical of credit cards and loans. They preferred cash, and they had canned goods stacked up everywhere. That’s just the way they were. They couldn’t help it.
My version of that was that I was at IBM when it was at its peak. It was the most admired company in the world. And within just a few years, it was nearly bankrupt. Hundreds of thousands of IBMers lost their jobs. Stockholder value was decimated, and the CEO was fired. It was humiliating and painful.
It makes you wonder—how in the world could a company that provided so much value to employees, shareholders, communities, and customers just completely lose it? It had every advantage in the world. It had money, market share, brand permission, respect. It could do anything. And it completely lost it.
So when I see a new technology emerge, I don’t dismiss it, nor do I say it’s the next great thing. You have to think about innovation very deeply, because the potential of misreading what it is and what it isn’t is tremendous.