When machines took over repetitive physical labor, many humans turned to knowledge-based work. But now machines can do that, too; a group of insurance workers was recently laid off in Japan, displaced by a computer system that can calculate payouts to policyholders. This new category of unemployment could have a serious impact on the economy.
That will be only one of many possible impacts of the deployment of artificial intelligence suggested by Bart Selman, professor of computer science, in a lecture, “The Future of AI: Benefits vs. Risks,” Feb. 27 in Olin Hall.
The talk was the kickoff lecture in a series on “The Emergence of Intelligent Machines: Challenges and Opportunities,” co-created by Selman and Joseph Halpern, professor of computer science, to be given on successive Monday evenings.
Selman began by reviewing milestones: IBM’s Deep Blue computer defeated chess champion Garry Kasparov. Watson won at Jeopardy, showing not only that it had a vast store of knowledge but that it can understand questions and formulate answers. In your pocket, Alexa and Siri are waiting to do the same. Along with speech, computers have mastered vision, with “superhuman” ability to recognize faces.
In another example of economic impact, self-driving cars now cruising around Southern California promise to put many transportation workers out of their jobs. Less obviously, because autonomous cars will have far fewer accidents, there may well be fewer openings for emergency-room medics. Which raises ethical and moral questions, Selman added: Should a self-driving car hit a pedestrian to protect the life of its passenger?
We may also see autonomous weapons, raising political issues. Will nations have to engage in nonproliferation treaties to control the development of self-guided drones?
There’s more on the way. Recent industrial investments in AI, Selman reported, come to well over a billion dollars. With hardware advances – an area he has studied – by the year 2035 you should be able to buy for about $1,000 a computer with the memory and processing capacity of the human brain.
To prepare for all this, Selman concluded, we should equip the machines to understand and pursue human values (another area of his research), and “ensure a tight coupling between human and machine interests.”
The next lecture in the series will feature Jon Kleinberg, the Tisch University Professor of Computer Science, on “Inherent Trade-offs in Algorithmic Fairness,” at 7:30 p.m. March 6, in 155 Olin Hall. The lecture series is part of a new computer science course, CS 4732 Ethical and Social Issues in AI.