WelcomeUser Guide
ToSPrivacyCanary
DonateBugsLicense

©2026 Poal.co

184

(post is archived)

The day we can teach computers how to think autonomously, making decisions based on past experiences to achieve the best results, regardless of hard-wired instructions: is the day I welcome our new overlords.

A little too soon. The day the AI can be more innovative and creative than humans for any given problems will be the day I will welcome the robot overlords. But by then, they’ll have to have the entire laws of the universe in their database anyway.

[–] 0 pt

How's that? Humans don't have all, most, or even many of the laws of the universe in their databases.

So you're agreeing that the complete AI takeover will take much longer than we anticipate? Because human innovation is basically combining the laws of the universe to give humans higher level of existence/consciousness. We already have the rudimentary understanding of the laws of physics, the question remains how much there is to know that'll determine how quickly the AI can take over. But my understanding certainly is that we don't know jack shit.

[–] 0 pt

So you're agreeing that the complete AI takeover will take much longer than we anticipate?

Depends on what you mean by "we". Following the the fastest supercomputers at top500.org I've calculated on three occasions years apart that in 2035 there will exist a computer capable of emulating a human brain entirely in CAD. On one hand we may find a better way, on the other hand we'll probably run into some difficulties no matter what the approach. That's before considering how much of the human mind is devoted to things that are already more efficiently done with computers and to maintaining a body.

So to be middle of the road between optimism and pessimism I think a strong AI will be built around 2035. But that AI will be buggy, incomplete, and require enormous amounts of electricity and other overhead. You need 7 more doublings (about 14 years) for that to shrink and mature down from what only a superpower can afford to what a small business can. So call that 2045 for computers to be well on their way to taking over complex administrative and other tasks requiring a sophisticated understanding of humans and the world, innovation, transfer of skills and awareness, human level communication, and so forth.

How long do you think it will take?