arb_

5 ms·

1. 1.▲
   What do we tell the humans?(theaidigest.org)

   3 points by arb_ 10mo ago|0 comments
2. 2.▲
   A new Moore's Law for AI agents(theaidigest.org)

   2 points by arb_ 1y ago|0 comments
3. 3.▲
   AIs are becoming more self-aware. Here's why that matters(theaidigest.org)

   3 points by arb_ 2y ago|0 comments
4. 4.▲
   by arb_ 2y ago

   I think they tweet prototypes and then don't ship most prototypes. Seems reasonable to me, otherwise you have instant bloat.
5. 5.▲
   GPT-4o web agent demo(theaidigest.org)

   3 points by arb_ 2y ago|0 comments
6. 6.▲
   by arb_ 2y ago

   As models improve, human preference will become worse as a proxy measurement (e.g. as model capabilities surpass the human's ability to judge correctness at a glance). This can be due to more raw capability - or more persuasion /
7. 7.▲
   by arb_ 2y ago

   Empirically, they have reduced hallucinations. Where do OpenAI / Anthropic claim that their models won't hallucinate?