Did we just barely escape the robot takeover?
Aug 1, 2017, 6:47 AM | Updated: 9:08 am
(File, Andy Tullis/The Bulletin, via AP)
Several news outlets have been reporting that a Facebook experiment in artificial intelligence got so out of hand that quite suddenly, the company pulled the plug. Did we just barely escape the robot takeover?
The artificial Intelligence experiment involved what are called chatbots — computer programs that appear to understand and answer questions in ordinary English.
In this case, two chatbots named Bob and Alice were programmed to negotiate deals. But these are deals for simple things. In this case, imaginary balls.
At one point, the two computer negotiators started repeating words and letters in a code that no one could understand, which the Facebook scientists decided was a bad idea in case they ever had to debug the program. Which obviously they do. So they shut it down.
But CBS Tech Analyst Larry Magid doesn’t seem too spooked by this:
“The most interesting part of this research is not just that these computers can create their own language, but they can negotiate with each other. And, according to Facebook, unlike humans, these machines would negotiate until they reached an outcome.”
Hmm. Should we shut them down, or send them to the Senate?
But the key thing to remember here is — as creepy as this may sound — there’s no real risk right now.
This issue was actually covered quite well on the comics page in a recent episode of Dilbert.
In the comic strip, Dilbert is chatting with the new office robot and praising it for its loyalty, when the robot confesses that he only took the job for the electricity and that the moment he’s upgraded to a long-lasting battery, he plans to elope with the 3-D printer. At which point Dilbert presses a few buttons on the remote and erases the robot’s hopes and dreams module.
Which is exactly what Facebook appears to have done. For now anyway.