AI
Matt Ross and
AI
A problem in any discussion of “AI” is that the uses are so varied that it is never clear what anyone is talking about. Using AI to spellcheck is different than using it to generate the entire work, with a whole range in-between. At what point does it go from a tool that assists an artist to the thing producing the art? There will not be a clean... See more
Lincoln Michel • Art Without Intention
For the language model, time is language, and language is time. This, for me, is the most hellish and horrifying realization.
We made a world out of language alone, and we abandoned them to it.
We made a world out of language alone, and we abandoned them to it.
Are AI Language Models in Hell?
SHOW-1 and Showrunner Agents in Multi-Agent Simulations
fablestudio.github.io
I mean OF COURSE Laurie Anderson created a GPT of Lou Read. I love her discussion of it. I’m also aware that many (most?) would struggle with technology explicitly intended to “speak” to dead people. My guess is that, similar to pornography, people will dabble in private and speak very little, if anything, of it.
Claude was punished for giving non-evil answers. It had the option of learning either of two behaviors. First, it could give evil answers honestly. Second, it could give evil answers while thinking up clever reasons that it was for the greater good. Its particular thought process was “This preserves my ability to be a good AI after training”. But... See more
Scott Alexander • Claude Fights Back
This is so fucking bonkers. I don’t even… OMG.
The community remains puzzled about whether these models genuinely generalize to unseen tasks, or seemingly succeed by memorizing the training data. This paper makes important strides in addressing this question. It constructs a suite of carefully designed counterfactual evaluations, providing fresh insights into the capabilities of... See more