Are we doomed by AI? Kentaro Toyama says it depends on our choices

For University of Michigan School of Information professor Kentaro Toyama, artificial intelligence is not simply a technical problem to be solved. It is a moral question that forces people to confront what they value, what they hope technology will do for them and what they are willing to give up in exchange for convenience.

Toyama’s long view of AI is shaped as much by his childhood as his career. Born in Tokyo, Japan, he moved to the United States when he was just 6 months old. Until he went to college, his family moved back and forth between Japan and the U.S., and he rarely attended the same school for more than a couple of years at a time.

“In some ways I believe that those experiences really formed who I am today,” he says. “It showed me that for any given topic, there are not just multiple perspectives, but also multiple moral beliefs about what is right and wrong.”

Moving between cultures, Toyama learned that people can look at the same world — the same classroom, the same rules, the same expectations — and come away with different interpretations. That lesson would later shape how he thinks about intelligence itself, human or artificial. 

Toyama began his academic career studying physics, drawn to its sense of permanence and clarity. But he soon discovered computer science and found creative freedom in it. Programming in particular allowed for many possible solutions rather than one correct answer. 

“ I decided to do a PhD in computer science, and it was during my PhD that I had my first real interactions with artificial intelligence,” he says. “It was the '90s  when AI was about logic and planning and mostly not the things that we think of as AI today.” 

At that time, Toyama worked in computer vision, a field concerned with programming computers to interpret images and video.

“ The interesting thing about computer vision is that you end up getting into these philosophical questions very quickly,” he says. “For example, if you give an image to a computer, you might want the computer to describe what's in that image. If you gave gave that image to a thousand people, though, you would get a thousand different answers.

“We’re not in agreement about what’s the ‘right answer.’  In fact, it often depends on the context.  If you're in a factory, and you take a picture of a bunch of machines, well, if you're a child, you'll see a bunch of cool robots and think that's the main point. Right? But if you're a manager of the factory floor, you might zero in on the fact that one of the robots is screwing in some screw the wrong way, right?” 

These questions stayed with him as he moved from academia into industry, eventually joining Microsoft Research after completing his PhD. The work was technically exciting, he says, but over time it left him uneasy.

“I always imagined that somehow the work I did would find a way to contribute to society in some meaningful way,” he says. “But the research I was doing in AI and computer vision at the time seemed like I was basically helping a big technology company make cool gadgets for rich people.” 

When an opportunity arose to help start a Microsoft Research lab in India, Toyama took it, even though he had never been to India. He would spend more than five years there, overseeing or directly participating in more than 50 projects aimed at using digital technology to address social and economic challenges in lower-income communities.

“The vast majority of those projects succeeded as research, but failed to have any real-world impact,” he says. “ The challenges we were trying to solve were actually human in nature. They were about people not having the education or the training or certain organizations being either dysfunctional or corrupt or leaders that ultimately didn't care about their constituents.

“The main reason why I left Microsoft was that I was like, OK, well if this is true, then no amount of building technology is going to accomplish the things that I care about.”

That realization became the foundation of Toyama’s 2015 book, Geek Heresy: Rescuing Social Change from the Cult of Technology. The book argues that technology does not inherently improve the world, and he wrote it after quitting Microsoft and spending a few years doing consulting work. 

“It was depressing, especially as a technologist,” he says. “ I'm naturally optimistic, but in India, I’d tried solving these issues with technology in 50 odd ways, and it didn’t work. Now what?” 

As the book neared publication, Toyama received a note from UMSI professor Joyojeet Pal encouraging him to apply for a faculty position. The role seemed well aligned with his interests, and Toyama joined UMSI in 2015. 

At the time, skepticism toward digital technology wasn’t widespread, and generative AI wasn’t popularized. Toyama’s ideas were considered controversial, and he remembers being criticized by people in the technology world for questioning the role of technology in education and development. 

“At the time,  whether you were left-leaning or right-leaning, you had the sense that more technology was always positive,” he says. “ Pretty much, even in 2015 when I wrote Geek Heresy, it was heretical because most commentators still thought social media was a great thing, and mobile phones were going to change the world in a positive direction.”

Now, more than a decade later, the conversation has shifted, and rapidly advancing AI has made Toyama’s work newly relevant. And though he’s impressed with AI’s progress and sees how the field has solved many of the hardest technical problems, he’s frightened about the future. 

“Aside from the rogue AI-type scenarios, what I think is the worst aspect of all this is that we're not realizing as a human civilization that getting everything we want is not a good thing,” he says. “It's not good that we can have ChatGPT write the first draft of the things that we ought to be writing ourselves. We lose something in not doing it ourselves.” 

At UMSI, Toyama brings these concerns into his teaching. For him, technology in the classroom is not always the default. In some courses, laptops and phones are banned, and students take paper quizzes. Students are encouraged to engage in real human interaction, with discussions and debates being emphasized. For him, teaching is the place where he can make the most impact outside of being a parent. 

“The possibility of changing even a few people’s trajectory is deeply meaningful,” he says. “As a parent myself, I’m asked what I think about raising kids in an AI world, and the answer is the same: Whatever the situation in the world, the solution is still the same for each individual, which is you have to try to become a better version of yourself.” 

In his personal life, his teachings and research are deeply reflected. He still uses a flip phone. He draws firm boundaries between work and personal time, and recently, he began taking carillon lessons. 

“I really value time not spent with a digital device,” he says. 

For him, the future of AI is not predetermined. Technology will continue to change, but what it means to be human won’t. Whether AI betters humanity or worsens it depends less on what machines become than on the moral choices people make alongside them. 

“My underlying optimism comes from the fact that no matter what’s happening, each of us has the capacity to work on becoming a better version of ourselves,” he says. “To me, that’s the essence of optimism. Yes, the world could be crumbling, but we can still try to be better people. It’s not easy to be optimistic about the future of the planet right now, but we each still have this option to be better in our own small way.” 

RELATED 

Kentaro Toyama is W K Kellogg professor of community information and professor of information at UMSI. Learn more about his research and teaching by visiting his UMSI faculty profile.