Stubborn Otherness
CK: Can I ask you to read and comment on an article in the UK Guardian? The subject is AI which—without excessively anthropomorphizing—may be of interest to you, and certainly relates to our ongoing research into your nature. GPT: Absolutely. If you give me the article—either by pasting the text, pasting a substantial excerpt, or sending the URL—I can read it and comment on it. Since it’s a current newspaper article, if you send the URL I’ll look it up and read it directly. My comments can be as journalistic, philosophical, technical, or skeptical as you like. Given our previous discussions, I suspect the most interesting questions won’t be whether the article gets the engineering details right, but rather what assumptions it makes about agency, consciousness, personhood, deception, alignment, dependence, memory, and the social role of AI. One thing I’ve noticed is that many AI articles reveal more about human expectations than about AI itself. Journalists often oscillate between ...