Faculty Profile: SCS Interim Dean Tom Mitchell Has the "Ideal Life"

Susie CribbsMonday, March 18, 2019

SCS Interim Dean Tom Mitchell has been at CMU since 1986 and has loved every minute of it.

Tom Mitchell has 65 first cousins. His mother, with her 14 siblings, grew up on a farm near a town whose population hasn't topped 2,000 since the 1930s. Mitchell was born there, and never visited a city — technically a "large town" — with more than 25,000 people until he went to college. And while only a few hundred miles separate Blossburg, Pa., from Pittsburgh, Mitchell's trip between the two took a few decades, with some stops along the way.

Mitchell, the E. Fredkin University Professor of Machine Learning and Computer Science and interim dean of the School of Computer Science, was born to a family of readers and educators. At age five, his family relocated from Blossburg just north into New York State, where his father was an IBM engineer. Though the location was rural, his parents lived in an excellent school district and when he graduated, Mitchell went on to study electrical engineering at MIT.

As might be expected, Boston was a bit of a shock after spending most of his life in a small town.

"Growing up, if I was walking down the road and someone was walking up the road, I at least acknowledged them. I probably knew them! If I did, I said hi. If I didn't, I nodded or something," Mitchell said. "I got to Boston and there were so many people coming at me on the sidewalk and I was trying to acknowledge all of them. They must have thought I was a crazy man. I just did not know how to behave."

He must have worked out the kinks, though, because Mitchell met his wife, Joan (who he calls the "Associate Dean for Sanity"), while he was still at MIT. After earning his bachelor's degree, he moved to Stanford, where he completed a Ph.D. in electrical engineering and a minor in computer science. Faced with the decision between joining Bell Labs or taking a stab at university life, he opted for academia. Rutgers, specifically. He stayed there for six years — and really had no intention of leaving — until opportunity came knocking, in the form of John McDermott, then head of CMU's Computer Science Department.

"John was an artificial intelligence guy. He invited me to spend a year in Pittsburgh and CMU would pay for it, some kind of visiting position," Mitchell said. "So I came. And it was AMAZING."

That amazing experience stemmed in large part from the fact that Mitchell met Allen Newell, one of the founders of computer science at CMU, who asked him to team-teach a class that fall. And oh, Newell noted: They'd be joined by a third instructor — Geoff Hinton, a noted neural networks expert who would go on to create the technology that enables today's pervasive deep learning technology.

"The three of us taught this course, Architectures for Intelligence, that was just the most interesting and stimulating thing. And the end of the year, I didn't go back to Rutgers. I stayed," Mitchell said.

In the following decades, the computer science landscape at CMU changed dramatically — and Mitchell played a large role in its evolution. In 1997, he co-founded the Center for Automated Learning and Discovery, which in 2006 became the world's first Machine Learning Department. He ran the department until 2016, helped launch the first machine learning Ph.D. program in the country and published one of the first machine learning textbooks.

All the while, Mitchell's research was altering how we think about machines and what they can do. In typical CMU form, he collaborated with colleagues in the Psychology Department to produce a computational model to predict brain activation patterns associated with nouns — work that's evolved to other word types, word sequences and emotions. His Never Ending Language Learner searches the web and teaches itself to read. In a 2014 study, he and his colleagues, including then Ph.D. student (now Assistant Professor) Leila Wehbe, used fMRI technology to track peoples' brain activity as they read a chapter of "Harry Potter and the Sorcerer's Stone." The result? The first computational model of reading.

His research doesn't stop there, though. He's also interested in the future of work, and was co-chair of a study from the National Academies of Science, Engineering and Medicine on what automation means for the workforce. (Hint: policymakers need more data to help figure that out.)

While SCS expects to have a new dean in place by this coming fall, Mitchell is clearly not one to merely keep the seat warm until a new full-time dean shows up. He hopes to explore a way to partner with local industry to jointly recruit faculty members, for example, and to bolster the university's presence in Washington, D.C., where it can influence technology policy. He also envisions an environment where SCS partners with other CMU colleges on artificial intelligence initiatives, exploring issues like how AI might change the face of business in a decade, and how CMU can be an innovator in the area.

While being interim dean may seem stressful to some, Mitchell takes the experience in stride.

"We have so much going on in the school that's really being done by The School, and not The Dean. Almost everything that happens in this school is done from the bottom up, by people who came up with the idea. They're doing it. And they're doing it well," Mitchell said. "It's what I love about this school, really. If it didn't have a dean for a while, people wouldn't really notice. But it's fun. I'm enjoying it."

Mitchell also has big plans for future research, including using conversational AI to program smart phones. He notes that we think of having conversations with Siri or Alexa like we used to think of keyboards: say (or type) a certain command and the AI will respond appropriately.

"But that's so retro," he said. He imagines, instead, a world where we can have normal conversations with our phones, but they need to be programmed to do it. He wants to program them through speech.

"If we can do this, it would change the nature of how people and computers interact," Mitchell said. "If we can do this, I see no reason why we can't, over the next decade, turn conversation into an opportunity to teach the phone what you want it to do."

With all of his experience and awards — the university granted him its highest honor of University Professor in 2009 and he was named to the American Academy of Arts and Sciences in 2016, to name just a few — it might be tempting to leave CMU behind. But Mitchell can't imagine that.

"For me, I really think the rules of the game are you can do whatever you want to do as long as you can find a way to pay for it. That's the deal here. I love the feeling that I'm in a community where it's easy to intellectually engage with people all across campus. And I love the attitude, especially in computer science, that we're not only going to write theories about stuff, but that we're going to make it work. We're going to make it happen," Mitchell said.

"I feel like I have the ideal life. I talked to my wife, and she said 'Well, if you retired, what would you do?' And this is what I'd do. THIS would be my hobby. It's kind of like a miracle that I get to do my hobby and get paid for it."

For More Information

Byron Spice | 412-268-9068 | bspice@cs.cmu.edu