Not really AI (happens every time) philipbrewer.micro.blog
@philipbrewer After looking at some of the eerily human-like behavior of LLMs, then reading that they're just giant autocomplete programs, I've thought more than once: Isn't that how a lot of our thinking and talking works? And isn't that why these programs look so human? Not something we'd like to admit.
@ReaderJohn @JohnBrady Yeah, ouch, that's sharp. So many "opinions" (including my own) are just repetitions of what others have said.
@tinyroofnail I might send this to my boss who is talking constantly about AI these days. But I'd have to run it through chatgpt so that it can be summarized into five or fewer bullet points before she'll read it.
@JohnBrady It is interesting. My mother-in-law suffered from dementia, and one result was that she'd think of something to say, start to say it, and then forget what she was trying to say. So she'd try to "figure it out" from the last few words she'd said, and veer off into some other story (from a steadily shrinking pool of stories). Which, now that LLMs exist, sounds rather a lot like what LLMs do (although their pool of stories aren't yet shrinking).