7 ms·

This has cosmological significance if it leads to superintelligence

by arisAlexis 2y ago

iLoveOncall 2y ago
Don't worry, it'll only lead to superstupidity.
- bluescrn 2y ago
  And superplagiarism of human-created content
  - XenophileJKO 2y ago
    I'm sure this will age well.
- _heimdall 2y ago
  Is that the prequel to Idiocracy?
Cthulhu_ 2y ago
It won't unless there's another (r)evolution in the underlying technology / science / algorithms, at this point scaling up just means they use bigger datasets or more iterations, but it's more finetuning and improving the existing output then coming up with a next generation / superintelligence.
- iLoveOncall 2y ago
  > bigger datasets Not even, they already ran out of data.
  - nick__m 2y ago
    I am sure that the M.I.C. have a ton of classified data that could be used to train a military AI.
- Filligree 2y ago
  Okay, but let’s be pessimistic for a moment. What can we do if that revolution does happen, and they’re close to AGI? I don’t believe the control problem is solved, but I’m not sure it would matter if it is.
  - ForHackernews 2y ago
    Being pessimistic, how come no human supergeniuses ever took over the world? Why didn't Leibniz make everyone else into his slaves? I don't even understand what the proposed mechanism for "rouge AI enslaves humanity" is. It's scifi (and not hard scifi) as far as I can see.
    - Philpax 2y ago
      Once you have one AGI, you can scale it to many AGI as long as you have the necessary compute. An AGI never needs to take breaks, can work non-stop on a problem, has access to all of the world's information simultaneously, and can interact with any system it's connected to. To put it simply, it could outcompete humanity on every metric that matters, especially given recent advancements in robotics.
      - ForHackernews 2y ago
        ...so it can think really hard all the time and come up with lots of great, devious evil ideas? Again, I wonder why no group of smart people with brilliant ideas has unilaterally imposed those ideas on the rest of humanity through sheer force of genius.
        Philpax 2y ago
        An equivalent advance in autonomous robotics would solve the force projection issue, if that's what you're getting at. I don't know if this will happen with any certainty, but the general idea of commoditising intelligence very much has the ability to tip the world order: every problem that can be tackled by throwing brainpower at it will be, and those advances will compound. Also, the question you're posing did happen: it was called the Manhattan Project.
- miki123211 2y ago
  > It won't unless there's another (r)evolution in the underlying technology / science I think reinforcement learning with little to no human feedback, O-1 / R-1 style, might be that revolution.
  - nkingsy 2y ago
    There is lots of human feedback. This isn’t a game with an end state that it can easily play against itself. It needs problems with known solutions, or realistic simulations. This is why people wonder if our own universe is a simulation for training an asi.
  - talldayo 2y ago
    I think gluing wings to a pig will make it fly. Show me examples or stop the conjecture.
computerthings 2y ago
"this generation shall not pass"... to me that's about as credible as wanting to "preserve human consciousness" by going to Mars. Setting the world on fire and disrupting societies gleefully, while basically building bunkers (figuratively more than literally) and consolidating surveillance and propaganda to ride out the cataclysm, that's what I'm seeing. And the stories to sell people on continuing to put up with that are not even good IMO. Just because the people who use the story to consolidate wealth and control are excited about that, we're somehow expected to be excited about the promise of a pair of socks made from barbed wire they gave us for Christmas. It's the narcissistic experience: "this is shit. this benefits you, not me. this hurts me." One thing is sure, actual intelligence, regardless of how you may define it, something that is able to reason and speak freely, is NOT what people who fire engineers for correcting them want. It's not about a sort of oracle for humanity to enjoy and benefit from, that just speaks "truth".

This has cosmological significance if it leads to superintelligence

iLoveOncall 2y ago

Don't worry, it'll only lead to superstupidity.

bluescrn 2y ago
And superplagiarism of human-created content
- XenophileJKO 2y ago
  I'm sure this will age well.
_heimdall 2y ago
Is that the prequel to Idiocracy?

Cthulhu_ 2y ago

It won't unless there's another (r)evolution in the underlying technology / science / algorithms, at this point scaling up just means they use bigger datasets or more iterations, but it's more finetuning and improving the existing output then coming up with a next generation / superintelligence.

iLoveOncall 2y ago
> bigger datasets Not even, they already ran out of data.
- nick__m 2y ago
  I am sure that the M.I.C. have a ton of classified data that could be used to train a military AI.
Filligree 2y ago
Okay, but let’s be pessimistic for a moment. What can we do if that revolution does happen, and they’re close to AGI? I don’t believe the control problem is solved, but I’m not sure it would matter if it is.
- ForHackernews 2y ago
  Being pessimistic, how come no human supergeniuses ever took over the world? Why didn't Leibniz make everyone else into his slaves? I don't even understand what the proposed mechanism for "rouge AI enslaves humanity" is. It's scifi (and not hard scifi) as far as I can see.
  - Philpax 2y ago
    Once you have one AGI, you can scale it to many AGI as long as you have the necessary compute. An AGI never needs to take breaks, can work non-stop on a problem, has access to all of the world's information simultaneously, and can interact with any system it's connected to. To put it simply, it could outcompete humanity on every metric that matters, especially given recent advancements in robotics.
    - ForHackernews 2y ago
      ...so it can think really hard all the time and come up with lots of great, devious evil ideas? Again, I wonder why no group of smart people with brilliant ideas has unilaterally imposed those ideas on the rest of humanity through sheer force of genius.
      - Philpax 2y ago
        An equivalent advance in autonomous robotics would solve the force projection issue, if that's what you're getting at. I don't know if this will happen with any certainty, but the general idea of commoditising intelligence very much has the ability to tip the world order: every problem that can be tackled by throwing brainpower at it will be, and those advances will compound. Also, the question you're posing did happen: it was called the Manhattan Project.
miki123211 2y ago
> It won't unless there's another (r)evolution in the underlying technology / science I think reinforcement learning with little to no human feedback, O-1 / R-1 style, might be that revolution.
- nkingsy 2y ago
  There is lots of human feedback. This isn’t a game with an end state that it can easily play against itself. It needs problems with known solutions, or realistic simulations. This is why people wonder if our own universe is a simulation for training an asi.
- talldayo 2y ago
  I think gluing wings to a pig will make it fly. Show me examples or stop the conjecture.

computerthings 2y ago

"this generation shall not pass"... to me that's about as credible as wanting to "preserve human consciousness" by going to Mars. Setting the world on fire and disrupting societies gleefully, while basically building bunkers (figuratively more than literally) and consolidating surveillance and propaganda to ride out the cataclysm, that's what I'm seeing. And the stories to sell people on continuing to put up with that are not even good IMO. Just because the people who use the story to consolidate wealth and control are excited about that, we're somehow expected to be excited about the promise of a pair of socks made from barbed wire they gave us for Christmas. It's the narcissistic experience: "this is shit. this benefits you, not me. this hurts me." One thing is sure, actual intelligence, regardless of how you may define it, something that is able to reason and speak freely, is NOT what people who fire engineers for correcting them want. It's not about a sort of oracle for humanity to enjoy and benefit from, that just speaks "truth".