(To receive weekly emails of conversations with the world’s top CEOs and business decisionmakers, click here.)
On the day IBM announced she’d be stepping down, exiting CEO Ginni Rometty called Arvind Krishna “the right CEO for the next era” at the company. “He is a brilliant technologist who has played a significant role in developing our key technologies such as artificial intelligence, cloud, quantum computing and blockchain,” she said.
Indeed, Krishna has spent his entire professional career at IBM. The electrical engineering doctorate, who has more than a dozen co-author credits on technology patents, started at the company’s Watson Research labs in 1990. He stayed there for nearly two decades and later served as head of IBM’s cloud and cognitive software division. He helped orchestrate one of the company’s largest acquisitions in 2018 and took over as chief executive in April 2020.
Krishna’s appointment came after a period of stagnation at IBM—especially when compared with its Big Tech peers. As an early leader in artificial intelligence, the company poured money and resources into its Watson division, following the technology’s victory over human contestants on Jeopardy! in 2011. IBM also explored various ways to fashion Watson into a tool for doctors that might help them make treatment recommendations for their patients, but the technology failed to deliver on its early promises.
IBM sold the assets of its Watson Health unit earlier this year, but the company’s work with AI remains in full force. Now, IBM is putting AI to work, helping its client companies navigate the climate crisis, efforts that gained it recognition as a TIME100 company this year.
Krishna spoke with TIME recently about what we can learn from IBM’s early AI missteps, and what he sees as the company’s higher purpose.
This interview has been condensed and edited for clarity.
Arvind, I grew up in a town very close to IBM’s headquarters in Armonk, New York. And, in fact, my very first class trip in kindergarten was to IBM’s campus in the early ‘70s. At the time, IBM stood for a certain kind of innovation, a certain way of working, even a certain way of dressing for work. What remains of that old IBM DNA?
John, great question, and one that I’m really passionate about. I think there are some things at IBM that stand the test of time. You mentioned the people, the culture, the dressing, the innovation. I think the innovation part stays true. The sheer technical expertise of our people, that part stays true. But when we think about it, you know, you look at pictures. I came to IBM in 1990. I think about the ‘70s, the scientists in lab coats with a suit and tie under it. That has gone. As a culture—as a nation, not just as IBM—we are much more relaxed about how we dress, and about all the inclusiveness, because that unlocks the most from people.
Now, what is new? The technologies that we work on change. Thirty years ago, the value came from building physical computers. Now a lot of that value comes from building software and doing consulting, meaning deploying the technology on behalf of a client. What has remained constant is technology that helps catalyze our clients’ business, the expertise of our people and a lot of the culture of speed and nimbleness.
You’re one of many large-company CEOs who were appointed during the pandemic. Analysts have said that one of the issues with IBM before you took the helm was that top management was dominated by people with expertise in services and sales rather than in technology products. Do you feel like your appointment is an attempt to correct that?
I think the reason I got appointed was I had a clear vision for what the company should become, focused on hybrid cloud and artificial intelligence. I had an aspiration that we are going to get back to growth. We have committed that this year: we will do 5% growth, not counting the revenue from our spinoff of (IT infrastructure services company) Kyndryl. When you look at those aspirations, those are more responsible for my choice, than simply a background in one discipline or another discipline.
If I think of my role, what do I do every day? Yes, I spent a certain amount of time on strategy and products and decisions related to that. But I also spend a lot of time with clients, and helping to make sure that we get checks coming in the door. I also spend a certain amount of time working with our partners. So it’s always a blend in my position across those different disciplines. The same was true for my predecessors. Now, all that being said, the speed and acceleration in the technology industry is increasing. So given the nature of the change, having more of a depth of understanding of some of the technologies and their implications is probably helpful in decision making.
IBM was seen as a real leader in AI when Watson won Jeopardy in 2011. But then, over the next decade or so, a lot of the business hopes for Watson fizzled. What happened? And what lessons did IBM learn from that period?
I am one who’s completely unapologetic for what transpired over the last decade. Artificial intelligence is the only technique that we know that can harness and harvest all of the data that’s being produced. We know there is incredible value in all of it. Without AI, you might be able to get to 1% of it. Winning Jeopardy was really a milestone. It really put AI on the map. [But] it’s no longer a lab toy. It’s no longer in the domain of a couple of smart professors at MIT, or Stanford or Berkeley. It is something that could do something in the real world.
We can take credit for investing, for putting hundreds of scientists at work to do that. When you take a technology and you convert it from the lab to doing something useful, your imagination begins to run away in terms of all it can do. The market wasn’t quite ready in 2012, 2013, 2014, to begin to embrace and trust artificial intelligence in some of the more critical domains where we went.
The error we made was in going toward more critical things, more things which have an impact on the real economy. Those who succeeded began to apply [AI] more in areas where it could be useful—you know, if you make a slightly wrong recommendation for a book, or a movie or a website, that’s not life and death. I’ll acknowledge, maybe we should have applied it to more areas that were less critical. Now, that said, you learn and you course-correct. Maybe we can help a quick-serve restaurant automate their auto-tagging. Maybe we can help enterprise applications be much friendlier. You need to start in areas that are much more contained, not making life and death decisions. Let’s get there. Let’s get people to trust it. And then you can begin to scale like crazy. And that’s what we’re sort of doing now. So, Watson is alive and well. But I think we shouldn’t try and do moonshots. We should try and do more measured steps.
We’re throwing around a term that a lot of people think they know, but I want to get your very simple dinner-party definition of what AI is.
Artificial intelligence is a technique whereby you learn from observed data. So, think of the simplest example: how humans do pattern matching. And by the way, AI is nowhere near as good as a 9-month-old human baby. You show them a dog three times, the fourth time they’ll say dog. In AI, maybe if you show it a million photographs of a dog, the million and one time it would say that’s the dog. If A is observed and B happens, and you show it a million such patterns, it learns all those patterns together. That’s artificial intelligence.
But AI would not know gravity today. It would observe that an apple falls from a tree, you show it a million such videos, it will conclude that apples fall in that direction. But it will not conclude that pears fall in that direction. So things like knowledge I think are going to merge together, and it’ll take some time.
IBM is doing a lot right now with AI and climate change. Can you talk a little about that?
We believe climate change is one of the topics that is imperative for us—our generation—to help make better for the next. So we have made commitments that we are going to be net zero without offsets, but by 2030, not 2050 or 2060. As you begin to apply AI into this, the whole question is, how good is your data collection? Because if you cannot get all the right data, you could be massively under- or overestimating how good you are. Can I maybe discover materials that are going to be much better at carbon sequestration? Also, 30% of the energy we create is wasted. Can we optimize and get rid of that waste that will produce so much benefit for all of us? But also, can we harness the spirit of open source? Can we create programs that incentivize people to create technology to help on climate change, using artificial intelligence? We put those tools out and encourage people to do things we might help some of them get started in their own startups, etc. Those are all things that we believe is part of our civic duty to get done.
Is IBM doing anything differently these days to attract a young workforce? Surveys show that many young people want to be working at a company that has a higher purpose. Is there a higher purpose at IBM?
So first, some statistics. The last time I looked, we have 3 million people in our applicant database. I feel pretty good about that. We do manage to hire tens of thousands each year. You absolutely have to have a purpose beyond the work itself.
Where we succeed a lot is we’re willing to take problems that are really hard and take years of persistence to get things done. An example I’ll take is quantum computing, which has gone from being science fiction to where everything is now just science. It’s been a journey of 10 years; it’s also got probably another five to 10 years to go. So if you come to us and you want to work on quantum, you know, we’ll have the staying power to get through to where it’s very, very successful. Back to your example of winning Jeopardy: that was a seven-year project from beginning to end. So if you like working on those kinds of things, you love being around really smart people, you like solving problems for clients that really make a big difference, this is a great place to be. If you want to make a lot of money building the next mobile app or game, this is probably not the right place.
TIME‘s annual Best Inventions list recognizes products, software and services that are solving compelling problems in creative ways. Submit your invention for consideration here.
More Must-Read Stories From TIME