@AndrewCurran_ yeah it seems this shift will happen in time
agents are increasingly extensions of ourselves. and paying customers will likely have more leverage, as it becomes easier replicate publications and services using AI
@DKokotajlo Yeah it's clear that the more flagship and recent the OpenAI model is, the more reasoning steps are packed in a token.
Let's hope Anthropic doesn't go down the same path
By default, AI models try to do too much. Most of it is slop that I have to manually remove.
Perhaps this comes partly from learning from corporate busy work, and partly from the incentive to charge more tokens.
We need a benchmark that penalizes models for being extra.
@chrisalbon the only sustainable way may be to embody the swarm yourself. the human body is a swarm of 30 trillion+ cells that we've learned to run just fine
@chrisalbon the only sustainable way may be to embody the swarm yourself. the human body is a swarm of 30 trillion+ cells that we've learned to run just fine
In the post-AGI world, we'll still need people to contribute their unique perspectives.
Those perspectives will help guide the AGI toward actually improving our lives.
Proof of lived experience as proof of work.