Positive news about AI
Peter Cameron
I have been, and remain, sceptical about AI. At best, it is saiad to be good at writing programs, and finding specific facts; but it has a tendency to lie, to invent, and to tell you what it thinks you want to hear, so you have to check very carefully everything it tells you.
But I was sent a manuscript by Donald Knuth where he claims that Anthropic’s robot Claude (named after Claude Shannon, the pioneer in information theory) has gone some way to revise his previously sceptical opinion. Claude.