Ranter
Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API

From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
Comments
-
kiki372773y@Demolishun we're talking about very distant future. CPU op/s has physical limit, same with architecture (can't go much lower than 5nm). But neural networks scale by cores, thus a physical limit (running out of silica) is extremely high. There is no other fundamental threshold of AI development.
-
kiki372773y@Demolishun I still want to see a self-supervised learning system of ten trillion artificial neurons in operation, I think it will be interesting to say the least.
-
So who will build that AI?
You didn't build the hamster and AI by evolution is what we are.
Advanced AI on advanced hardware will be as bad as humans - but way faster and without any conscience. One slightly misformulated rule and AI will happily sacrifice whole humanity for a minuscule metric increase. Think stereotypical entrepreneur - but with a thousand times the recklessness. -
AI is a fancy form of applied statistics. It has nothing to do with thinking or understanding a problem.
-
arcioneo7683y@kiki
Not sure if it was you, did you posted something like this anywhere?
“I like to blow my partner, to see that huge strong person melting in my mouth…”
Something like that, am searching for that text -
kiki372773y@Oktokolo evolution in computer systems are trivially done by simulating genetic drift and natural selection. That's the only two concepts you need to make things evolve. We already have those tools and the whole paradigm of evolutionary programming. Also, remember that in computer systems one generation doesn't need to last years. Few seconds per generation will be enough.
-
@kiki: So humans will select the generations that will survive. What you will get is what i warned about.
-
kiki372773y@Oktokolo if we have a rigorous mathematical framework that allows comparing random mutations to figure out what's "better", that's all we need to do. As soon as it remains maths, there is no room for opinion. There is no room to be human, and no possibility to find out if that framework was created by a human at all
-
@kiki: But life has no inherent goal and good or evil isn't something you can describe mathematically. If you try to base ethics on math, you get Stalin - basically the insane AI from the Paranoia role-playing game:
https://en.wikipedia.org/wiki/..._(role-playing_game) -
kiki372773y@Demolishun Professor Dave had an amazing video on origins of life, I really suggest it. Ignore all James Tour stuff because drama.
https://yewtu.be/watch/...
https://yewtu.be/watch/... -
TheEnd6883yThis is a great and fun discussion, thanks for bringing up the topic. I came across this the other day that I found very interesting: https://nature.com/articles/...
You can comprehend its whole construction completely in two seconds. Yet, a hamster will be entertained by exploring this thing for life.
In the same way, an advanced neural network will be able to figure out our brain's construction and explain it to us.
If you cry AI takeover, remember that just because you can kill a hamster with your hand, and it absolutely can't do anything about it, doesn't mean you'll do this.
Said neural network may have morals completely detached not only from ours, but from the whole concept of "morals" as we know it. Its goals being beyond our understanding doesn't mean it will be hostile and won't help us.
The only thing we'll lose is control. Yet, benefits are so huge that they can transfer us up within the Kardashev scale, and it may be our only way to prevent the death of our civilization.
We don't have control over our nature either. We can't prevent eruptions and earthquakes. Losing control in itself doesn't mean the thing we lost control on will kill us.
random