@AdithyaNLP Wow, really cool!
I feel like the most obvious application of this, is an incredibly good chess teaching app, where the moves by high-end engines are no longer mysterious to humans.
Do you have any ideas or thoughts on this?
@mhmazur I am curious about the human baseline measurement:
Were the humans told and incentivized to achieve the goal in as little actions as possible?
If they weren't, could it be that their action efficiency would be a lot higher, if they had been?
Kinda like a "human max" setting?
@fchollet How realistic do you think it is we will succeed at this?
Given all the incentives in the real world to remove "slow humans" from the loop, it sounds extremely difficult?
My fool proof plan to contribute something useful to this world...
Be the first person to fire off this prompt to GPT 11:
Solve all of humanities problems. Make no mistakes.
@anilkseth@dwarkesh_sp@OpenAI@huggingface Um, he of course doesn't think the agents feel anything or have subjective experience... He just used those anthropomorphisms to get his points across quicker.