[HN Gopher] Don't believe the hype: AGI is far from inevitable
       ___________________________________________________________________
        
       Don't believe the hype: AGI is far from inevitable
        
       Author : mpweiher
       Score  : 42 points
       Date   : 2024-09-29 19:02 UTC (3 hours ago)
        
 (HTM) web link (www.ru.nl)
 (TXT) w3m dump (www.ru.nl)
        
       | loa_in_ wrote:
       | AGI is about as far away as it was two decades ago. Language
       | models are merely a dent, and probably will be the precursor to a
       | natural language interface to the thing.
        
         | lumost wrote:
         | It's useful to consider the rise of computer graphics and cgi.
         | When you first see CGI, you might think that the software is
         | useful for _general_ simulations of physical systems. The
         | reality is that it only provides a thin facsimile.
         | 
         | Real simulation software has always been separate from computer
         | graphics.
        
         | Closi wrote:
         | We are clearly closer than 20 years ago - o1 is an order of
         | magnitude closer than anything in the mid-2000s.
         | 
         | Also I would think most people would consider AGI science
         | fiction in 2004 - now we consider it a technical possibility
         | which demonstrates a huge change.
        
       | sharadov wrote:
       | The current LLMs are just good at parroting, and even that is
       | sometimes unbelievably bad.
       | 
       | We still have barely scratched the surface of how the brain truly
       | works.
       | 
       | I will start worrying about AGI when that is completely figured
       | out.
        
         | diob wrote:
         | No need to worry about AGI until the LLMs are writing their own
         | source.
        
       | pzo wrote:
       | So what? Current LLM has been really useful and can be still
       | improved to be used in million robots that need to be good enough
       | to support many specialized but repetitive tasks - this would
       | have tremendous impact on economy itself.
        
       | Gehinnn wrote:
       | Basically the linked article argues like this:
       | 
       | > That's because cognition, or the ability to observe, learn and
       | gain new insight, is incredibly hard to replicate through AI on
       | the scale that it occurs in the human brain.
       | 
       | (no other more substantial arguments were given)
       | 
       | I'm also very skeptical on seeing AGI soon, but LLMs do solve
       | problems that people thought were extremely difficult to solve
       | ten years ago.
        
         | babyshake wrote:
         | It's possible we see some ways in which AI becomes increasingly
         | AGI like in some ways but not in others. For example, AI that
         | can create novel scientific discoveries but can't make a song
         | as good as your favorite musician who creates a strong
         | emotional effect with their music.
        
           | KoolKat23 wrote:
           | This I'm very sure will be the case, but everyone will still
           | move the goalposts and look past the fact that different
           | humans have different strengths and weaknesses too. A tone
           | deaf human for instance.
        
         | godelski wrote:
         | > but LLMs do solve problems that people thought were extremely
         | difficult to solve ten years ago.
         | 
         | Well for something to be G or I you need them to solve novel
         | problems. These things have interested most of the Internet and
         | I've yet to see a "reasoning" disentangle memorization from
         | reasoning. Memorization doesn't mean they aren't useful (not
         | sure why this was ever conflated since... Computers are
         | useful...), but it's very different from G or I. And remember
         | that these tools are trained for human preferential output. If
         | humans prefer things to look like reasoning then that's what
         | they optimize. [0]
         | 
         | Sure, maybe your cousin Throckmorton is dumb but that's besides
         | the point.
         | 
         | That said, I see no reason human level cognition is impossible.
         | We're not magic. We're machines that follow the laws of
         | physics. ML systems may be far from capturing what goes on in
         | these computers, but that doesn't mean magic exists.
         | 
         | [0] If it walks like a duck, quacks like a duck, and swims like
         | a duck, and looks like a duck it's _probably_ a duck. But
         | probably doesn 't mean it isn't a well made animatronic. We
         | have those too and they'll convince many humans they are ducks.
         | But that doesn't change what's inside. The subtly matters.
        
           | stroupwaffle wrote:
           | I think it will be an organoid brain bio-machine. We can
           | already grow organs--just need to grow a brain and connect it
           | to a machine.
        
             | Moosdijk wrote:
             | The keyword being "just".
        
               | ggm wrote:
               | Just grow, just connect, just sustain, just avoid the
               | many pitfalls. Indeed just is key
        
               | godelski wrote:
               | just adverb        to turn a complex thing into magic
               | with a simple wave of the hands            E.g. To turn
               | lead into gold you _just_ need to remove 3 protons
        
             | idle_zealot wrote:
             | Somehow I doubt that organic cells (structures optimized
             | for independent operation and reproduction, then adapted to
             | work semi-cooperatively) resemble optimal compute fabric
             | for cognition. By that same token I doubt that optimal
             | compute fabric for cognition resembles GPUs or CPUs as we
             | understand them today. I would expect whatever this
             | efficient design is to be extremely unlikely to occur
             | naturally, structurally, and involve some very exotic
             | manufactured materials.
        
           | danaris wrote:
           | I have seen far, far too many people say things along the
           | lines of "Sure, LLMs currently don't seem to be good at
           | [thing LLMs are, at least as of now, _fundamentally incapable
           | of_ ], but hey, some people are pretty bad at that sometimes
           | too!"
           | 
           | It demonstrates such a complete misunderstanding of the basic
           | nature of the problem that I am left baffled that some of
           | these people claim to actually be in the machine-learning
           | field themselves.
           | 
           | How can you not understand the difference between "humans are
           | _not absolutely perfect or reliable_ at this task " and "LLMs
           | _by their very nature_ cannot perform this task "?
           | 
           | I do not know if AGI is possible. Honestly, I'd love to
           | believe that it is. However, it has not remotely been
           | demonstrated that it is possible, and as such, it follows
           | that it cannot have been demonstrated that it is inevitable.
           | If you want to believe that it is inevitable, then I have no
           | quarrel with you; if you want to _preach_ that it is
           | inevitable, and draw specious inferences to  "prove" it, then
           | I have a big quarrel with you.
        
         | tptacek wrote:
         | Are you talking about the press release that the story on HN
         | currently links to, or the paper that press release is about?
         | The paper (I'm not vouching for it; I just skimmed it) appears
         | to reduce AGI to a theoretical computational model, and then
         | supplies a proof that it's not solvable in polynomial time.
        
           | Gehinnn wrote:
           | I was referring to the press release article. I also looked
           | at the paper now, and to me their presented proof looked more
           | like a technicality than a new insight.
           | 
           | If it's not solvable in polynomial time, how did nature solve
           | it in a couple of million years?
        
             | jprete wrote:
             | Nature is not necessarily computational; the size of the
             | problem might also be such that four _billion_ years of
             | evolution was enough.
        
       | ngruhn wrote:
       | > There will never be enough computing power to create AGI using
       | machine learning that can do the same [as the human brain],
       | because we'd run out of natural resources long before we'd even
       | get close
       | 
       | I don't understand how people can so confidently make claims like
       | this. We might underestimate how difficult AGI is, but come on?!
        
         | fabian2k wrote:
         | I don't think the people saying that AGI is happening in the
         | near future know what would be necessary to achieve it. Neither
         | do the AGI skeptics, we simply don't understand this area well
         | enough.
         | 
         | Evolution created intelligence and consciousness. This means
         | that it is clearly possible for us to do the same. Doesn't mean
         | that simply scaling LLMs could ever achieve it.
        
           | nox101 wrote:
           | I'm just going by the title. If the title was, "Don't believe
           | the hype, LLMs will not achieve AGI" then I might agree. If
           | it was "Don't believe the hype, AGIs is 100s of years away"
           | I'd consider the arguments. But, given brains exist, it does
           | seem inevitable that we will eventually create something that
           | replicates it even if we have to simulate every atom to do
           | it. And once we do, it certainly seem inevitable that we'll
           | have AGI because unlike brain we can make our copy bigger,
           | faster, and/or copy it. We can give it access to more info
           | faster and more inputs.
        
             | threeseed wrote:
             | > it does seem inevitable that we will eventually create
             | something
             | 
             | Birds fly. Therefore it seems inevitable that humans will
             | evolve to also fly.
             | 
             | Also don't forget that many suspect the brain may be using
             | quantum mechanics so you will need to fully understand and
             | document that field. Whilst of course you are simulating
             | every atom in the universe using humanity's _complete_
             | understanding of _every_ physical and mathematical model.
        
           | umvi wrote:
           | > Evolution created intelligence and consciousness
           | 
           | This is not provable, it an assumption. Religious people
           | (which account for a large percent the population) claim
           | intelligence and/or consciousness stem from a "spirit" which
           | existed before birth and will continue to exist after death.
           | Also unprovable, by the way.
           | 
           | I think your foundational assertion would have to be
           | rephrased as "Assuming things like God/spirits don't exist,
           | AGI must be possible because we are AGI agents" in order to
           | be true
        
         | staunton wrote:
         | For some people, "never" means something like "I wouldn't know
         | how, so surely not by next year, and probably not even in ten".
        
         | chpatrick wrote:
         | "There will never be enough computing power to compute the
         | motion of the planets because we can't build a planet."
        
         | Terr_ wrote:
         | I think their qualifier "using machine learning" is doing a lot
         | of heavy lifting here in terms of what it implies about the
         | engineering approach, cost of material, energy usage, etc.
         | 
         | In contrast, imagine the scenario of AGI using artificial but
         | _biological_ neurons.
        
       | tptacek wrote:
       | This is a press release for a paper (a common thing university
       | departments do) and we'd be better off with the paper itself as
       | the story link:
       | 
       | https://link.springer.com/article/10.1007/s42113-024-00217-5
        
       | Gehinnn wrote:
       | I skimmed through the paper and couldn't make much sense of it.
       | In particular, I don't understand how their results don't imply
       | that human-level intelligence can't exist.
       | 
       | After all, earth could be understood as solar powered super
       | computer, that took a couple of million years to produce
       | humanity.
        
       | 29athrowaway wrote:
       | AGI is not required to transform society or create a mess beyond
       | no return.
        
       | gqcwwjtg wrote:
       | This is silly. They article talks like we have any idea at all
       | how efficient machine learning can be. As I remember it, the LLM
       | boom came from transformers turning out to scale a lot better
       | than anyone expected, so I'm not sure why something similar
       | couldn't happen again.
        
         | fnordpiglet wrote:
         | It's less about efficiency and more about continued improvement
         | with increased scale. I wouldn't call self attention based
         | transformers particularly efficient. And afaik we've not hit
         | performance with increased scale degradation even at these
         | enormous scales.
         | 
         | However I would note that I in principle agree that we aren't
         | on the path to a human like intelligence because the difference
         | between directed cognition (or however you want to characterize
         | current LLMs or other AI) and awareness is extreme. We don't
         | really understand even abstractly what awareness actually is
         | because it's impossible to interrogate unlike expressive
         | language, logic, even art. It's far from obvious to me that we
         | can use language or other outputs of our intelligent awareness
         | to produce awareness, or even if goal based agents cobbling
         | together AI techniques is even approximate to awareness.
         | 
         | I suspect we will end up creating an amazing tool that has its
         | own form of intelligence but will fundamentally not be like
         | aware intelligence we are familiar with in humans and other
         | animals. But this is all theorizing on my part as a
         | professional practitioner in this field.
        
       | avazhi wrote:
       | "unlikely to ever come to fruition" is more baseless than
       | suggesting AGI is imminent.
       | 
       | I'm not an AGI optimist myself, but I'd be very surprised if a
       | time traveller told me that mankind won't have AGI by, say, 2250.
        
       | coolThingsFirst wrote:
       | Zero evidence given on why it's impossible.
        
       | SonOfLilit wrote:
       | > 'If you have a conversation with someone, you might recall
       | something you said fifteen minutes before. Or a year before. Or
       | that someone else explained to you half your life ago. Any such
       | knowledge might be crucial to advancing the conversation you're
       | having. People do that seamlessly', explains van Rooij.
       | 
       | Surprisingly, they seem to be attacking the only element of human
       | cognition that LLMs already surpassed us at.
        
         | azinman2 wrote:
         | They do not learn new facts instantly in a way that can rewrite
         | old rules or even larger principals of logic. For example, if I
         | showed you evidence right now that you were actually adopted
         | (assuming previously you thought you werent), it would rock
         | your world and you'd instantly change everything and doubt so
         | much. Then when anything related to your family comes up this
         | tiny but impactful fact would bleed into all of it. LLMs have
         | no such ability.
         | 
         | This is similar to learning a new skill (the G part). I could
         | give you a new tv and show you a remote that's unlike any
         | you've used before. You could likely learn it quickly and
         | seamlessly adapt this new tool, as well as generalize its usage
         | onto other new devices.
         | 
         | LLMs cannot do such things.
        
       ___________________________________________________________________
       (page generated 2024-09-29 23:00 UTC)