Superficially resembling cognition =/= cognition. I'm quite excited about many of the specific use cases for LLMs, and have worked a few things into my own met

Superficially resembling cognition =/= cognition. I'm quite excited about many of the specific use cases for LLMs, and have worked a few things into my own methods of doing things. It's a quick and convenient way to do lots of actual specific things. For example: if I want to reflect on different ways to approach a (simple) maths problem, or what sorts of intuitions lie behind an equation, it is helpful to have a tool that can sift through the many snippets of text out there that have touched off that and similar problems, and present me with readable sentences summing up some of those snippets of text from all those places. You've to be very wary, as highlighted by the article, but as "dumb summarisers" that save you trawling through several blogs, they can be quicker to use. Nonetheless, equating this with "reasoning" and "intelligence" is only possible for a field of academics and professionals who are very poorly versed in the humanities. I understand that tech is quite an insular bubble, and that it feels like "the only game in town" to many of its practitioners. But I must admit that I think it's very possible that the levels of madness we're witnessing here from the true believers will be viewed with even more disdain than "blockchain" is viewed now, after the dust has settled years later. Blockchain claimed it was going to revolutionise finance, and thereby upend the relationship between individuals and states. AI people claim they're going to revolutionise biology, and life itself, and introduce superintelligences that will inevitably alter the universe itself in a way we've no control over. The danger isn't "AI", the danger is the myopia of the tech industry at large, and its pharaonic figureheads, who continue to feed the general public - and particularly the tech crowd - sci-fi fairytales, as they vie for power.