Post a response:
|
I guess GPT admitting its lies is something. I tried getting it to translate a particularly rude piece of rhetoric from Clement of Alexandria, and it was giving me obviously bogus summaries, too short and mild with no hint of fulmination. I would quiz GPT, and it would admit it was a summary, not a translation, then proceed to give me a translation which was virtually the same reduced summary. Weirdly, it was censoring a denunciation of the vile sin of sodomy, at that time primarily a pederastic affair. It will be a very peaceful land, living as Eloi under AI's beneficent eye. I've also seen discussions by folk keen on the science of IQ and its heritability. AI, when asked questions on the science, will outright lie and say there's no evidence for heritability of IQ. Which is a lie. Called on this, AI will apologize and actually start quoting the science that exists. And it doesn't even blush! As long as one is familiar with the answers one seeks, AI is just great. GLTTA would make up part of Chat GPT's data store - it must cause the poor chap no end of algorithmic indigestion! |