“By 2029 no computer - or "machine intelligence" - will have passed the Turing Test.” This is the $20,000 bet made in 2002 between Mitch Kapor (betting for true) and Ray Kurzwile (betting for false). In case, like me, you are not sure of the double negative, it is Ray that thinks a computer will pass the Turing test and exceed human general intelligence by 2029.
In her new book, Melanie Mitchell goes into this wager in detail. Spoiler alert: she is with Kapor: computers will not pass the Turing test by 2029. The book is excellent and worth reading. My very short summary is that computers are going to get very very good at the things they are naturally good at, but may never equal humans in general intelligence -- no matter how many doublings we get with Moore’s law. We under appreciate human general intelligence. A four year old human can make inferences that we may never see artificial intelligence achieve. Computers will get better and faster at translating spoken commands into an action, but may never be able to discern if a person is happy or a threat.
Mitchell brings some rational thinking back into the prognosticating about the advances in AI. She does not think we will see a computer reading and understanding War and Peace or programming themselves to overthrow their masters anytime soon.
While everyone talks about the computers taking over all the jobs, it is easy to forget that six million Americans lose or leave their jobs every month. And when times are good, a few more than six million people get new jobs every month. It is the net change that gets reported in the media like this:
Total nonfarm payroll employment rose by 225,000 in January, and the unemployment rate was little changed at 3.6 percent, the U.S. Bureau of Labor Statistics reported today. Notable job gains occurred in construction, in health care, and in transportation and warehousing.
There is rarely a mention of job turnover rate or the math that 25 times more people changed jobs last month than the net change in employment number. 160 million people have jobs in America so at a turnover rate of six million a month it takes a bit over two years for the entire workforce to turn over. Sure, some people hold onto their jobs for many more years, and some people change jobs every few months, but the point is, the US economy is extremely dynamic, and will adapt even if computers or robots take some of the jobs.
Daniel Suarez, in his 2009 book Daemon, introduces glasses that can see through walls. Unlike superman’s x ray vision however, the glasses are connected to a network that has access to the security cameras. In Suarez’s world, no one needs to actually see through the walls because there is a camera on the other side of the wall streaming the video. Just as useful but not in the way we expected.
Change is on the way. It is likely the change we are expecting will take a different form than we expected. We never got the flying cars or the lives of leisure predicted in the ‘50s. But we did get computers in our pockets that can beat us at chess, call us a cab, book us a flight, and execute complex stock trades.
If you take the time to read Artificial Intelligence: A Guide for Thinking Humans by Melanie Mitchell you will be better equipped to see what shape the changes are taking.
