Facebook Panics Pulls AI Lab Plug
Imagine a future world where computers are driving our cars everywhere for us and taking care of the mundane routines of our lives such as mowing the yard or preparing our meals. At some point sooner rather than later in order for them to function in the same environment without bumping into one another they'll have to communicate with each other. They'll have their own language that humans will develop, program into code and install into each machine.
So what if over time the computers began to develop their own little language 'short-cuts', their own slang if you will to accomplish the jobs we give them. By doing so they'd self-improve and do our work more efficiently and in less time wouldn't they. And that'd be great for us wouldn't it?
Eventually they'd advance from language shortcuts to more advanced communications skills to talk to each other. And in time they'd be able to invent their own language, one which we don't understand. An ability to communicate with each other completely foreign to us.
Could an enlightened machine with such 'artificial intelligence' and ability to evolve beyond language beyond our ability to comprehend ever develop a 'personality'?
Imagine your lawnmower became "aware" that over the years you've neglected its routine maintenance, causing its parts to wear unevenly and sooner than they should have causing un-necessary damage to it and a type of mechanical 'pain', could you enter your garage and face those vindictive whirling blades with an attitude coming straight for you?
It sounds so Stephen King'ish doesn't it?
Just a few days ago Tesla CEO Elon Musk said that artificial intelligence is our biggest risk and we should be 'very afraid'; and now it was reported that a Facebook experiment with AI has ran amok and had to be shut down immediately. They yanked the plug because they couldn't just tell it to stop. It wouldn't listen.
Facebook was forced to shut down one of its AI systems after it was discovered that chat-bots had started to communicate in a language they could not understand, and refused requests to switch back to English only. The full story was reported over the weekend in Tech Times.
What Facebook is working on is AI chat-bots to talk to you if you ever need assistance from the online Goliath. You're on your end thinking you're chatting with a lady named Veronica but it's really only bits and bytes. Veronica is virtual. You're responding to tiny bits of code that use AI to be able to shift their questioning to solve your problem.
But the Facebook bots developed their own language that the researchers couldn't turn off or control. Frankenchatbots.
But who is to say the chat-bots were getting nasty? Or would be rude to customers or take over the world and enslave us humans? What if artificial intelligence developed in a benevolent manner and the Facebook chat-bots were pre-disposed to start sending us checks in the mail? As you can imagine, Zuckerberg would want something like that shut down at once!
We're still some 18 years short of 2035 but Isaac Asimov would have to be proud of the this turn of events as he'd envisioned in his writings.