Don't listen to David Karpfs of the world. Did he predict chat gpt? if you asked him in 2018, he would have said AI will never write a story
now you can use AI to easily write the type of articles he produces and he's pissed.
now you can use AI to easily write the type of articles he produces and he's pissed.
You really cannot.
Seriously. This is just the parrot thing again. The fact that AI proponents confuse the form of words with authorial intent is mindbending to me.
Wouldn't have confused Magritte, I think.
Words are words. Writers are writers. Writers are not words.
ETA: consider what would actually be necessary to prove me wrong. And when you hear back from David Karpf about his willingness to take part in that experiment, write a blog post about it and any results, post it to HN.
I am sure people here will happily suggest topics for the articles. I, for example, would love to hear what your hypothetical ChatKarpf has to say about influences from his childhood that David Karpf has never written about, or things he believed at age five that aren't true and how that affects his writing now.
Do you see what I mean? These aren't even particularly forced examples: writers draw on private stuff, internal thoughts, internal contradictions, all the time, consciously and unconsciously.
"Words are words. Writers are writers. Writers are not words."
I'm very bullish on AI/LLMs but I think we do need to have a better shared understanding of what they are and what they aren't. I think there's a lot of confusion around this.
Thank you. I don't think it really explains the distinction, of course. It just makes it clear there necessarily must be one, and it can't be wished away by discussions of larger training sets, more token context, or whatever. It never will be wished away.
When I’m writing out a comment, there’s no muse in my head singing the words to me. I have a model of who I am and what I believe - if I weren’t religious I might say I am that model - and I type things out by picking the words which that guy would say in response to the input I read.
(The model isn’t a transformer-based LLM, of course.)
Come on.