@Greatnorthroad2 In case I haven’t spammed you with this already… twitter.com/chapmanbookman…

RT @gethistories: Histories: Falling, 1871 gethistories.com/p/falling-1871…

@robinhouston Yeah, hard to say. I suppose at the moment I think LLMs have inherent limitations that may not be fixable by just chucking processing power at them. But of course other solutions could emerge, and I’m definitely following with interest!

@robinhouston Can we have ‘fascinated and sceptical in equal measure’? I’m using it a lot but it’s got many flaws (or, rather, humans have).

I’m sure I can’t be the first to make the ‘putting the sot into Brightwell-cum-Sotwell’ joke today.

So I had a nice chat with ChatGPT about its own limitations and in fairness it was pretty open about them. The problem is really humans overinterpreting what it actually does (a form of pareidolia which @garymarcus has drawn attention to). Clever and fascinating, but not ‘AI’.

@richardf @Sustrans @NationalByway Bravo, sir!