2023 Alignment Research Updates from FAR AI 2023-12-04T22:32:19.842Z

What's new at FAR AI 2023-12-04T21:18:03.951Z

Even Superhuman Go AIs Have Surprising Failure Modes 2023-07-20T17:31:35.814Z

AI Safety in a World of Vulnerable Machine Learning Systems 2023-03-08T02:40:43.139Z

Deconfusing Landauer's Principle 2022-05-27T17:58:23.353Z

Comment by EuanMcLean (euanmclean) on Deconfusing Landauer's Principle · 2022-06-23T19:27:26.096Z · LW · GW

Yea, I think you're hitting on a weird duality between setting and erasing here. I think I agree that setting is more fundamental than erasing. I suppose when talking about energy expenditure of computation, each set bit must be erased in the long run, so they're interchangeable in that sense.

Comment by EuanMcLean (euanmclean) on Deconfusing Landauer's Principle · 2022-06-06T19:07:38.342Z · LW · GW

Sorry for the delay. As both you and TheMcDouglas have mentioned; yea, this relies on $H(C|X) = 0$. The way I've worded it above is somewhere between misleading and wrong, have modified. Thanks for pointing this out!

Comment by EuanMcLean (euanmclean) on Deconfusing Landauer's Principle · 2022-05-29T11:53:05.616Z · LW · GW

Fixed, thanks!

Comment by EuanMcLean (euanmclean) on Deconfusing Landauer's Principle · 2022-05-28T11:29:52.935Z · LW · GW

Thanks for the comment, this is indeed an important component! I've added a couple of sentences pointing in this direction.

Comment by EuanMcLean (euanmclean) on Deconfusing Landauer's Principle · 2022-05-28T09:46:09.939Z · LW · GW

fixed, thanks!