How AI Knows Things No One Told It – Scientific American

No one yet knows how ChatGPT and its artificial intelligence cousins will transform the world, and one reason is that no one really knows what goes on inside them. Some of these systems abilities go far beyond what they were trained to do and even their inventors are baffled as to why. A growing number of tests suggest these AI systems develop internal models of the real world, much as our own brain does, though the machines technique is different.

via www.scientificamerican.com

One of the YouTube videos I watched (and I watch many) had an AI big boffin referring to current AI LLM research as the “summoning and taming” approach. Step one, you “summon” a mind from out of the large language model soup, and Step two, you (try to) tame it. Not exactly reassuring. This guy — don’t remember his name, but he was one of the founders or first employees at Skype — was definitely in favor of a moratorium. I’m not. I don’t think it will do any good, except perhaps give Elon a chance to catch up a bit. But something weird is going on inside the LLM black box.