News for hackers. Stuff that matters.

A comment on RLHF is just barely RL. Parent comment

Re:RLHF is just barely RL by djeastm on Thursday August 08, 2024 @07:51AM (#41190488)
>Coding AI can write tests, write code, compile, examine failed test cases, search for different coding solutions that satisfy more test cases or rewrite the tests, all in an unsupervised loop.

Will this be able to be done without spending absurd amounts of energy?

[ Parent | on HN ]
Re:RLHF is just barely RL by jimbokun on Thursday August 08, 2024 @08:24AM (#41190795)
Energy efficiency might end up being the final remaining axis on which biological brains surpass manufactured ones before the singularity.
[ Parent | on HN ]
Re:RLHF is just barely RL by commodoreboxer on Thursday August 08, 2024 @09:49AM (#41191479)
The amount of energy is truly absurd. I dont chug a 16 oz bottle of water every time I answer a question.
[ Parent | on HN ]
Re:RLHF is just barely RL by Tostino on Friday August 09, 2024 @08:57AM (#41201395)
Neither do these models. The calculations I saw claiming some absurdly high energy or water use seemed like an absolute joke. Par for the course for a journalist at this point.
[ Parent | on HN ]
Re:RLHF is just barely RL by ben_w on Thursday August 08, 2024 @10:05AM (#41191681)
Computer energy efficiency is not as constrained as minimum feature size, it's still doubling every 2.6 years or so.

Even if they were, a human-quality AI that runs at human-speed for x10 our body's calorie requirements in electricity, would still (at electricity prices of USD 0.1/kWh) undercut workers earning the UN abject poverty threshold.

[ Parent | on HN ]