On AI

A conversation with friends

C.K.S. – There’s nowt any of us can do to stop every Tom, Dick, Harry & AI crawling anything we put on public access on the Internet. I don’t even try to slow the AI bastards down. A bit sad at my fiction getting nicked, but if my wisdom (haha) influences ChatGPT or DeepSeek, well, maybe they’ll be a picoscopic tad wiser… Last thing in the world I’d worry about is your crawler, P.F. Anyone who does is (imnsho) nuts. Crawl to your heart’s content.

S.P. – “A bit sad at my fiction getting nicked, but if my wisdom (haha) influences ChatGPT or DeepSeek, well, maybe they’ll be a picoscopic tad wiser…”

I came to the conclusion a while back that, at worst, these things are indiscriminate information gatherers; factual accuracy being the same as the totality of the Internet i.e. 50% +/- 50%

At best, the most accurate label would probably be AIS (Artificial Idiot Savant)

C.K.S. – “factual accuracy being the same as the totality of the Internet i.e. 50% +/- 50%”

This is why I’m unimpressed by attempts to censor the Internet: 50% +/- 50% is significantly better than TV or the papers, which are more like 33% +/- 33% … (That is, I prefer Accidental Idiocy to Assiduous Indoctrination…)

S.P. – C.K.S. My estimations of the percentage of accurate info are mostly likely off, but my queries do tend to generate a better percentage than many of my colleagues, and I don’t participate in Facepalm, Whitter, Instagraphic etc. so those are missing from the negative side of my estimate.

C.K.S. – “My estimations of the percentage of accurate info are mostly likely off”

I think 50% +/- 50% is probably pretty accurate for the Internet as a whole. Apply a reasonable (natural) intelligence filter and you can improve on that quite a bit. You can improve on the 33% +/- 33% of the Old Media with a good natural intelligence filter, too, but the missing 34% isn’t there for the filter to allow through, so 66% is an unreachable asymptotic limit.