Eric Horvitz receives ACM-AAAI Allen Newell Award for groundbreaking artificial intelligence work

Apr 27, 2016   |   Allison Linn

In his many years as an artificial intelligence researcher, Eric Horvitz has worked on everything from systems that help determine what’s funny or surprising to those that know when to help us remember what we need to do at work.

On Wednesday, Horvitz, a technical fellow and managing director of Microsoft’s Redmond, Washington, research lab, received the ACM – AAAI Allen Newell Award for groundbreaking contributions in artificial intelligence and human-computer interaction. The award honors Horvitz’s substantial theoretical efforts and as well as his persistent focus on using those discoveries as the basis for practical applications that make our lives easier and more productive.

Harry Shum, the executive vice president of Microsoft’s technology and research group, said Horvitz epitomizes a style of research that is unique to places like Microsoft because it is focused on having an impact in both the research and industry domains.

“People talk about basic research and applied research. What we are doing here is Microsoft research,” Shum said. “It’s not just about doing theoretical research and writing more papers. It’s also about applying those technologies in Microsoft products.”

Jeannette M. Wing, the corporate vice president overseeing Microsoft’s core research labs, said that Horvitz’s research has had an impact on countless research projects and commercial products, ranging from systems that help make our commutes easier to ones that seek to prevent hospital readmissions.

“His impact is immeasurable,” she said.

But Wing noted that Horvitz also has been able to step back and see the big picture, becoming a visionary and a thought leader in a field that is growing increasingly complex.

“He asks big questions: How do our minds work? What computational principles and architectures underlie thinking and intelligent behavior? How can computational models perform amidst real-world complexities such as sustainability and development? How can we deploy computation systems that deliver value to people and society?” Wing said.

The Newell award is given to a researcher whose work has breadth within computer science or spans multiple disciplines. Horvitz’s work has combined multiple computer science disciplines and he has been a leader in exploring the interrelationships between artificial intelligence and fields like decision science, cognitive science and neuroscience.

The award comes at a time when the artificial intelligence field is exploding.

Until a few years ago, artificial intelligence wasn’t often part of the public consciousness, except when it came up in a science fiction novel or blockbuster movie.

Now,  thanks to breakthroughs in the availability of data and our ability to process it, artificial intelligence applications are  suddenly everywhere, including systems that can understand and translate language, recognize and caption photos and do increasingly smart and useful things for us.

During a time often referred to as the “AI winter,”  Horvitz was among the nation’s hard-charging researchers plugging away at the difficult work of laying the groundwork for these systems and thinking about how they would work in the real world. Although artificial intelligence was out of the spotlight during that time, researchers were making major breakthroughs in bringing together the logical methods of traditional artificial intelligence work with research in fields such as decision science. This led to new applications that used both logic and probability.

Horvitz said that many of his research projects over the last fifteen years – which have looked at things like what we are most likely to remember or forget and when it’s worth it to interrupt someone while working – foreshadow practical applications that he expects to see in the future.

“To me, Eric is such an epic example of those brilliant researchers who have this huge confidence — not over-confidence, but just confidence — to keep pushing forward,” Shum said.

Horvitz’s attention to both research advances and practical applications of artificial intelligence research began while he was pursuing his Ph.D. on principles of bounded rationality. That’s the idea that when people or computers make decisions, they are limited by time, available information and their reasoning abilities.

Horvitz said he was interested in how computing systems immersed in the real world could make the best decisions in time-critical situations. His research looked at the value of continuing to think about a problem versus stopping early with a good enough answer.

His research considered emergency room scenarios, in which artificial intelligence systems could help doctors with timely recommendations. The work foreshadowed his later research on using similar ideas to guide solutions to some of the hardest challenges known in artificial intelligence, in the realm of theorem proving.

Horvitz also showed how artificial intelligence systems could be used to better understand people’s goals and intentions and provide the best information to decision makers. He collaborated with NASA’s Mission Control Center on how to provide flight engineers with the most valuable information about space shuttle systems when the engineers are under intense time pressure.

To solve these problems — and many more after — Horvitz brought together artificial intelligence methods with ideas drawn from disciplines like probability theory, decision theory and studies of bounded rationality.

In the future, Horvitz said he sees vast possibilities for how artificial intelligence can help to augment human intelligence.

“There’s a huge opportunity ahead in building systems that work closely with people to help them to achieve their goals,” Horvitz said.

Related:

Tags: , , ,