Discussion
AGI is here.
Eufrat: No, it is isn’t and saying it is here by using vague goalposts does not make AGI show up. I will agree we have been unable to define what it is, but we can’t even figure out why humans have consciousness right now.
crooked-v: Let's see any of this stuff do something as simple as run a vending machine without collapsing into incoherency before calling it 'AGI'.
fastball: [delayed]
ripped_britches: I’m not on either side of the argument, but one popular definition is missing which is “can automate most knowledge work”.Not that this is my definition or anything, just pointing out that this is the one people actually care about, even if the acronym doesn’t say anything about economics or social change.
jfeew: We are yet to see customer service reps be displaced by functional agents - I’m not referring to half baked AI chats on uber eats that are designed to piss you off and go away.
jfeew: Why do people keep referencing the Turing test? Turing did not anticipate there’d be a gigantic dump of text contributed by humans online to feed off.
mapontosevenths: People mention it to remind the world that the goal posts have been repeatedly moved by critics, and always will be.A certain percentage of humans will never acknowledge that machines can be intelligent. Those people should be disqualified from the conversation for the same reason we disqualify biblical literalists from conversations about radio carbon dating.
oakhan3: I think we agree - we have arbritrary goalposts regarding AGI and they have been met. WE don't know what we consider to be "the big changing moment" and that moment is hard to define because we don't have a good definition of it when we talk about ourselves even.So the convo becomes - what is that "thing" and do we need to draw similarities between "it" and our own intelligence.
ipnon: The singularity doesn't mean that we get an AGI Day with a big announcement from The People In Charge that intelligence has been solved once and for all. It simply means that the frequency of "this changes everything" and "rumored model X at lab Y passes benchmark Z at 110% in less-than-zero-shot" style posts increases monotonically for ever.
rl3: The clickbait and ensuing arguments will never end. That is our fate.Once ASI exists we'll still have people arguing whether it's actually AGI or not.
atulvi: My definition has always been making a new world changing discovery in physics or math.
jfeew: That’s a limiting definition imo.
mapontosevenths: I have never met a human that managed that, and we still let them vote!
jfeew: This is nonsense.The big changing moment is when there is a clear trend of job destruction in the economy that can be directly explained by llm’s displacing labour.
jfeew: Intelligence is not simply the ability to recall information. That’s an aspect but not the entirety.BTW your attitude stinks.
ranger_danger: > A certain percentage of humans will never acknowledge that machines can be intelligent.Doesn't this assume there IS an objective, quantifiable definition of an "intelligent machine" that is agreed upon by most people? That instead sounds rather subjective to my ears.
oakhan3: Interesting, could you explain it further - I'd like to know what that means - I thought I covered this in definition 8 - but with the very blatant asterisk in the post being - we have it and it works but it doesn't work consistently enough and with acceptable quality in long enough runs or open ended tasks - but I believe we will get closer and closer to this target with improvements in models and scaffolding.
lolz404: Can confirm an easy way to win an argument is to remove the decenting voice. Bet that would have went great in civil right and woman sufferage.
brg: [delayed]
SpicyLemonZest: That seems like a pretty dangerous place to set your goalposts. Don't you think we need to see what's coming and figure out how to deal with it before widespread job destruction starts?
oakhan3: It is quoted relentlessly and so is worth addressing, for example see https://www.ibm.com/think/topics/artificial-general-intellig...I mentioned it to have a more complete set of definitions for AGI from across the community - but do agree that it is by far the weakest and more-so a measurement of human variability and gullibility than AI intelligence.
SpicyLemonZest: Or it will just become seen as a category error. We don't often go around talking about "artificial general strength" or "artificial super-strength", even though it's easy to build a claw arm which can exert more force in any direction than any human can.
ranger_danger: > we can’t even figure out why humans have consciousness right now.My uneducated guess is that it just means we save/remember (in a lossy way) inputs from our senses and then constantly decide what to do right now based on current and historical inputs, as well as contemplated future events.I think the rest of our body greatly influences all of that as well, for example: we know running is healthy and we should do it, but we also decide not to run if we are busy, feel tired, or are in pain etc.
mapontosevenths: Even failing a single unified definition, every reasonable person should be able to define some subjective line of their own.Some people don't even have a subjective definition though. They'll continue to deny the machines are intelligent no matter where the line is drawn.It's not worth debating those folks because to them it is a matter of faith and no amount of reason can convince the unreasonable.
oakhan3: That vending machine comment is probably referring to https://www.anthropic.com/research/project-vend-1 - which is a fair point.But I think it actually supports my thesis: we either haven't defined AGI well enough, or we've met it and are now waiting for something beyond it that doesn't have a name yet - something with the common sense and situational intuition a human would have in a scenario like that. The goalposts keep moving because the definition was never solid to begin with.
mapontosevenths: Ignoring the irrational isn't the silencing of dissent, it's ignoring time-wasters who refuse (or are unable) to argue in good faith.I only get so many hours on earth, I'd rather not spend them debating what the definition of "is" is with someone who would rather litigate tautological nonsense than accept *any* level of evidence as sufficient.
PaulDavisThe1st: At one point in time, "AGI" included being able to learn skills that involved manipulating the physical world. While LLMs and their ilk may contribute to this, we are still (AFAICT) far, far from this at this time.
PaulDavisThe1st: "the hard problem of consciousness" is not about "what are we conscious of", but rather: how is it possible to be conscious (i.e. experience qualia) at all?
PaulDavisThe1st: Why would it specifically be job-displacement-via-LLM's that signal AGI? Why not job-displacement-via-automated-robot? or job-displacement-via-office-technology?
polotics: AGI is here? Yann Le Cun has a few weeks ago once more presented his PoV about how current LLMs fail: https://youtu.be/nqDHPpKha_A?is=sQsO57UWwR8LGZkWin french ...so in my own words:1) Still unreliable at logic and general inference: try and try again seems to be SoTA...2) Comically bad at pro-activity and taking the right initiative: eg. "You're right to be upset."3) Most likely already reaching the end of the line in terms of available good training data: looking at the posted article here, I would tend to agree...
Petersipoi: Since when? I've never understood AGI to require that.
readthenotes1: Seems kinda arbitrary.More humans publish non-replicable "science" in sociology -- your bar is way too high
polotics: I think you will like Robert Sapolski lectures on YouTube...
NitpickLawyer: The problem is that LeCun was obviously wrong on LLMs before. You have to take what he says with the caveat that he probably talks about these in a purist (academic) way. Most of the "downsides" and "failures" are not really happening in the real world, or if they happen, they're eventually fixed / improved.~2 years ago he made 3 statements that he considered failures at the time, and he was quite adamant that they were real problems:1. LLMs can't do math2. LLMs can't plan3. (autoregressive) LLMs can't maintain a long session because errors compound as you generate more tokens.ALL of these were obviously overcome by the industry, and today we have experts in their field using them for heavy, hard math (Tao, Knuth, etc), anyone who's used a coding agent can tell you that they can indeed plan and follow that plan, edit the plan and generally complete the plan, and the long session stuff is again obvious (agentic systems often remain useful at >100k ctx length).So yeah, I really hope one of Yann, Ilya or Fei-Fei can come up with something better than transformers, but take anything they say with a grain of salt until they do. They often speak on more abstract, academic downsides, not necessarily what we see in practice. And don't dismiss the amout of money and brainpower going into making LLMs useful, even if from an academic pov it seems like we're bashing a square peg into a round hole. If it fits, it fits...
