Post #1897723

@jnsgruk I first encountered reports about this that put the LLM stuff forward and I reacted with revulsion, then read this link and my reaction was considerably more moderate. We really need some terminological clarity in this field that separates what I sometimes think of as "heuristic computing" (NNs, DL, NLP, expert systems &c; so TTS, speech/image recognition, &c) and stuff that's termed "generative AI" ("skeumorphic" models of various nature what get touted as having cognition and creativity somehow, ChatGPT, stable diffusion, Sora, Claude, &c). I am all for the former. Don't use them much myself but people do have valid use cases for it. I dislike the latter but I can live with it so long as it's some normally unobtrusive developer tool you can install/enable if you want. So long as it doesn't Microsoft Copilot its way into my life I'll hold my nose about it lol Lastly, and I mean this most constructively: I think your dichotomy of explicit/implicit AI is basically what I say above but those terms are confusing. Mine aren't better either. "AI" is polarising and tainted. I think it would be opportune to workshop the communication of this as a lot of people who aren't informed about these nuances are trying out or using Linux because they're fed up of "AI" in Windows in particular and this might generate unnecessary and avoidable confusion. In fact it seems to be doing that already.