The Economist magazine of Jne 25th to July 1st, 2016, contains a special report on Artificial Intelligence. It is entitled The Return of the Machinery Question, referring to the major discussions that took place during what we call the industrial revolution and concerned the effects of mechanization on society_ in particular employment for the working class. It is well worth reading, as we will all be affected, and the first part of the report is here.
When I joined the Stanford faculty in 1966, I spent some time with the Stanford Artificial Intelligence Laboratory (SAIL), led by John McCarthy and involved in what was later recognized as pioneering work. Since I was fresh from the Jet Propulsion Laboratory, and had been involved in the design of space craft, I was trying to define a research area for myself. My Ph.D. research, having to do with the remote control of vehicles, had moved to SAIL, and the new field of artificial intelligence was attracting a good deal of attention. It was not to be my main field of interest, since I have never been as excited about computers and software as maybe I should have been, but I have fond memories of my time there and the people involved.
At the time, SAIL had a couple of Digital Equipment Corporation PDP-8 computers wired together, relatively advanced at the time, but of course primitive by present standards. And the goals of the work were even more advanced, as has always been the case with the field of Artificial Iintelligence. And the work in which I was involved had to do with the control of machines. As time has passed, both digital hardware and software have advanced greatly, as has the amount of information available, but as the Economist report states, until recently the accomplishments in the field have seemed to be not as spectacular as might be expected.
But now, the field seems to be exploding, and as the report points out, impacting activities as varied as “improving internet search engines, blocking spam e-mails, suggesting e-mail replies, translating web pages, recognizing voice commands, detecting credit-card fraud, and controlling self-driving cars”. Rather than being disappointed in the progress in artificial intelligence, we seem to be becoming worried about its potential to replace people in large areas of employment. Some studies have shown that some 50% of all jobs are threatened by computed automation —white collar as well as blue.
The cause of this is a combination of radically increased computer power, new approaches to the use of computers, huge amounts of information becoming available, and large investments being made in the area. Such techniques as deep learning, which allow computers to use this information to find new patterns, develop new techniques for solving problems, and improve their own software are showing up in a wide variety of fields, from medicine to finance to education. And computers are increasingly beating humans at their own games —a recent example being DeepMind’s Alpha Go system’s defeat of champion Lee Sedol at Go, a board game considered much more complex than Chess.
The report in The Economist not only reviews the progress of Artificial Intelligence, but also discusses the possible effects of this progress on society. Here it is optimistic, relying on historical data that new developments in technology have always resulted in more new jobs than the developments have obsoleted. But the optimism is a bit offset by the emphasis on the power of these new developments, and by reference to such things as experiments in universal domestic income in Finland and the Netherlands in case they might be needed to keep economies going in the fact of major job loss—also by references to the unprecedented reach of computers as opposed to the manufacturing of the industrial revolution.
Life is never boring.