On cars, old, new and future; science & technology; vintage airplanes, computer flight simulation of them; Sherlockiana; our English language; travel; and other stuff
ARTIFICIAL INTELLIGENCE bots are learning to chat with each other in their own language. Using deep-learning skills to screw on bottle caps is one thing, but language is decidedly more complex. As Adrienne LaFrance observed in The Atlantic, June 15, 2017, “… the fact that machines will make up their own non-human ways of conversing is an astonishing reminder of just how little we know, even when people are the ones designing these systems.”
In an investor call on April 27, 2016, reported in The Verge, Mark Zuckerberg said, “… the biggest thing that we’re focused on with artificial intelligence is building computer services that have better perception than people, basic human senses like seeing, hearing, language, core things that we do. I think it’s possible to get to the point in the next five to 10 years where we have computer systems that are better than people at each of those things.”
For language, make that less than 15 months.
Researchers Shut Down AI That Invented Its Own Language is the headline at the Digital Journal website, July 21, 2017. Notes author James Walker, “The observations made at Facebook are the latest in a long line of similar cases. In each instance, an AI being monitored by humans has diverged from its training in English to develop its own language. The resulting phrases appear to be nonsensical gibberish to humans but contain semantic meaning when interpreted by AI ‘agents.’ ”
At the root of all this is deep learning, with the AI agents working beyond their initial human programming.
The Facebook AI bots are our old pals Alice and Bob, recast from their traditional quantum world. At Facebook Artificial Intelligence Research (FAIR), AI negotiators Alice and Bob began their computerized interactions in English.
But, before long, FAIR researchers monitoring the pair encountered Bob saying, “I can I I everything else,” to which Alice responded, “Balls have zero to me to me to me to me to me to me to me to.” The negotiation continued with each understanding what the other was saying.
Noted one of the human researchers, “There was no reward to sticking to English language.… Agents will drift off understandable language and invent codewords for themselves.”
Philosophical insights appear at CO.Design website’s article “AI is Inventing Languages Humans Can’t Understand. Should We Stop It?” by Mark Wilson. He notes that these languages are akin to the jargon that develops, for example, among pork belly traders at the Mercantile Exchange.
Wilson notes, “… it offers us the possibility of a more interoperable world, a more perfect place where iPhones talk to refrigerators that talk to your car without a second thought. The tradeoff is that we, as humanity, would have no clue what those machines were actually saying to one another.”
Should we discourage this covert communication?
“But at the same time,” Wilson observes, “it feels short-sighted, doesn’t it? If we can build software that can speak to other software more efficiently, shouldn’t we use that?”
Alice and Bob may have the last word. Even if we have no idea what they’re saying. ds
© Dennis Simanaitis, SimanaitisSays.com, 2017