22
US urged to consider military strikes to stop China achieving AGI first (www.scmp.com)
submitted 2 months ago by [B] to c/china@news.abolish.capital
11 comments fedilink hide all child comments
 

The United States should start preparing for scenarios where extreme measures must be taken to stop China from achieving artificial general intelligence (AGI), according to a former White House official, including state-backed espionage and military strikes on Chinese data centres. Jacob Stokes, deputy director of the Indo-Pacific Security Program at the Centre for a New American Security (CNAS), said at an online event on Thursday that various US agencies, including the Department of Defense...

From China - South China Morning Post via This RSS Feed.

you are viewing a single comment's thread
view the rest of the comments
[–] -4 points 2 months ago* (5 children)

Can you explain why you hold this viewpoint? From my perspective, all the tools are in place. Self adjusting weights, adaptive transformers, etc to actually achieve agi with enough compute and refinement. I'll be upfront that I'm only working with large abstract concepts, not the direct math, but I'm genuinely curious what makes you believe it's outside the scope of the first species to split an atom? Is there some formula or proven constant I'm missing?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 5 child comments
    •
  • [–] 5 points 2 months ago

    LLMs and generally deep learning as a whole have a number of serious limitations that would seemingly prevent AGI. That's not even getting into the fact that a lot of the fundamentals of LLMs havent changed since they were rolled out almost a decade ago, so it seems unlikely we struck gold in 2017 and just havent worked it out yet.

    A great example is that LLMs can't do most math on their own. The Dag like structure of deep learning puts a finite limit on the number of calculations that are possible.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 4 points 2 months ago (3 children)

    It's an efficiency problem, our brain work on 1/100000000 the energy required by a data center at a 1/100000 the size, it should tell you how efficient brains are.

    You can throw resources at it but you'll be limited by bandwidth.

    But the fundamental problem is that we don't know how consciousness works. And if the goal is to reach AGI, at what scope? Have it create new technology? Wouldn't those billions have been better invested in research teams?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] -2 points 2 months ago* (2 children)

    You could say the same thing about the computer in your hands 40 years ago dude, and why have humans in a life of unnecessary toil if we can accomplish all the day to day via machines? I just don't think I've got the same (somewhat justified) ai hate as most of lemmy.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 3 points 2 months ago

    it’s not just that either. the current implementation of “ai” is not artifical intelegence. these llm’s do not conceptualize the subject matter they regurgitate. by the very way they operate they can not achieve this. examples of this failure have already been demonstrated with the full glass of wine scenario, and the strawberry scrnario.

    they operate on the tokenization of data, and their relationships, but where as a mind can form relationships between these nodes, an llm must be directed to do so. these directions are what they scrape. were it able to conceptualize there would be no need to train the llm’s as throughly as they do.

    it’s why we get pizza with the cheese glued on top.

    llm’s have their uses. but what is being sold is snake oil in order to cause a bubble for profit. if llms were just fed information with in a very specific scope, it could be a very useful tool without the need for it to scale up anywhere near this badly

  • permalink
  • fedilink
  • source
  • parent
  • [–] 1 point 2 months ago

    Machine learning and neural network have their place but they are custom made for one task. Current "AI" or llms and image generators have very few real use cases and does not justify their resource use plus they make money for mentally ill persons.

    Also the unnecessary toil is very necessary, but that's philosophy.

    Anyway due to some quantum laws there is a hard barrier on how small you can make transistors or else electrons will just pass through, and we're getting close to it, so we can't reach AGI that way

  • permalink
  • fedilink
  • source
  • parent