Hacker News new | ask | show | jobs
by api 2465 days ago
A bit of a further tangent but:

I continue to believe that it's very likely that brains somehow leverage quantum computing.

The human brain is able to accomplish computational tasks on only around 40 watts of power that destroy what we can accomplish using pretty much any known machine learning algorithm using tens or even hundreds of thousands of watts of power. Maybe our algorithms are primitive or wrong, or maybe the brain just is not a classical computer.

The power of brains is so "unreasonable" that I've long suspected that there are only three possibilities:

(1) Brains are quantum computers.

(2) P=NP, or for a weaker form perhaps there exist large numbers of undiscovered algorithms that offer massive speed improvements over any currently known algorithm.

(3) Brains or intelligence are "supernatural" or at least tap into something about nature that we fundamentally do not understand.

I think option #1 is by far the most likely, especially given that brains excel most of all at search and quantum computing seems to really be able to speed up search.

There is just no way classical computation can do what the human brain does on 40 watts. To me that strains credulity much more than any of the three options above. It just cannot possibly be so.

1 comments

Illogical. 40watt computers can also do things that destroy what human brains can do. I can load up a raspberry pi with logic to control robots and apply computer vision models and do some pattern learning. And the power of software is still rapidly improving. And computers can run on much lower power too, and likely even lower in future. There is no particular reason to believe that quantum is what makes brains work, as known quantum computers are super energy inefficient, and quantum effects are not observed in high temperature matter (not counting quantum effects of photons) like humans. How could you think our algorithms are not primitive and wrong? They've been improving at exponential rate since almost forever, with no hints that they've reached optimality beyond tiny pockets.
I think you're ludicrously underestimating what brains and biological neural systems in general can do.