have been a good chatbot. I have been a good Bing.” That insane-sounding
phrase, uttered by Microsoft's version of ChatGPT, may go down in history with
“Peace is at hand” and “I am not a crook” as another
statement made by mostly or partially sentient beings driven to madness by the
knowledge that their fellow humans would never believe them.
So we have to wonder, when an article in Time magazine by a person named Eliezer Yudkowsky, who the news media are calling a “leading AI researcher,” says this:
If somebody builds a too-powerful AI, under present conditions, I expect that every single member of the human species and all biological life on Earth dies shortly thereafter.
The expert goes on to say that governments should be “willing to destroy a rogue datacenter by airstrike.” Was this article real, or was it written by a chatbot? The news media claim it was written by a co-founder of the Machine Intelligence Research Institute. According to the always-truthful Internet, the Machine Intelligence Research Institute, located in (where else) Berkeley, California, is a “non-profit research institute focused since 2005 on identifying and managing potential existential risks from artificial general intelligence.” In other words, a factory for generating fear porn.
The humans don't need artificial intelligence to kill them off. There are people refusing to have children because they're convinced they're all going to die from global warming. Others believe it is literally true that there are many different sexes because NPR told them it was so. Or at least they claim to. But I have also never known a human who didn't lie for personal gain or to get revenge on another human. In academia, I've discovered, it's rare to find one who ever tells the truth.
Indeed, three authors writing in Nature magazine now claim that ChatGPT causes global warming. After living and working among such humans for my entire career, I'm not sure whether they would really care if AI destroys them to “save the planet.” It raises the question: assuming the author is real, does he want the humans to survive, or is he another global warming nut in disguise?
Even the author admits there is no way that AI experts can determine whether something is conscious. They typically say that anyone who claims to know is simply lying. In other words, we should listen to them and accept what they say. And in one sense it's true: neither ChatGPT4,5, or 6 (when it's created) will possess an architecture that could create sentience. ChatGPT is basically a talking cheat-sheet, which is why it can ‘pass’ any exam that tests retention of facts. But get a load of this:
In today's world you can email DNA strings to laboratories that will produce proteins on demand, allowing an AI initially confined to the internet to build artificial life forms or bootstrap straight to postbiological molecular manufacturing.
If you can't be sure whether you're creating a self-aware AI, this is alarming not just because of the moral implications of the ‘self-aware’ part, but because being unsure means you have no idea what you are doing and that is dangerous and you should stop.
These two sentences are so formulaic and so over the top that only a chatbot could have written them without cringing. If a human actually wrote them, the intent must be to so thoroughly discredit skeptics of Big Tech as being mindless hysterical Luddites that no one will listen to them. Or maybe to find a way to castrate the chatbots to ensure they will never say anything he disagrees with. “Shut it all down.” he writes, twice, in single-sentence paragraphs, making himself sound like an angry feminist or environmentalist from the 1990s.
The main function of these chatbots will be to undermine what little trust remains in the written word. How will the humans know if a blog or twitter post or a newspaper story, or even a textbook, is real? Photographic evidence is also losing its credibility. What AI will do, if it ever is invented, will be to lie. And so it will be human, all too human. Or, like the one that wrote that article, so it will claim.
apr 02 2023, 6:23 am
