Remix.run Logo
▲ GalahiSimtam 6 hours ago

I am old enough to remember what AGI meant before the rise of LLMs, and it used to be more or less what this article calls "ASI".

You might find more people believing that AGI is already solved. For example, the author of "The AGI chronicles" drops this in the epilogue. For him, AGI is achieved, because it can help him with a wide range of subjects, with the ability of a random stranger. It's clear that he means chatbots, internet surveys summaries, and coding bots - he does not even mention self-driving cars there.

This is a shift from the earlier chapters, where AGI term is employed somewhat more vaguely, and even quipping once that AI researchers think of AGI as a slightly smarter version of themselves.

I too, have a lower bar for AGI. It would be enough for me if it could design games I work on. Merely speeding up implementation tasks is not that interesting. In the immortal words, a delayed game is eventually good, but a rushed game is forever bad.

▲lordnacho an hour ago | parent | next [-]

> Merely speeding up implementation tasks is not that interesting.

This allows you design things faster, though. More effort can be spent on iteration when you aren't fiddling with libraries and syntax. I find that I am able to discover the open questions a lot faster because you reach the decision point sooner when the machine is simply doing the trivial things.

As for the design aspects, it is pretty good at brainstorming. Whatever your choice is, I find that it throws out an exhaustive vanilla list for whatever it is. It can also critique whatever you come up with.

So for me, I also feel like AGI is already here. There's probably a few things it can't do as well as a real random person, but there's a heck of a lot of things that it does better.

▲stef25 9 minutes ago | parent [-]

> This allows you design things faster, though.

There will be no more Netflix shows to watch while Claude works for me, if it goes even faster.

▲jaapz 5 hours ago | parent | prev | next [-]

> but a rushed game is forever bad.

Is it though? CDPR famously released Cyberpunk rushed and in a miserable state. But the game is amazing now, and this is a widely held opinion too.

▲dracotomes 4 hours ago | parent | next [-]

I played Cyberpunk on release and replayed like a good half of it last year. I see this sentiment a lot online but it doesn't resonate. For me it still feels like the same empty beautiful world populated with same 50-70 NPCs that spawn in just around the corner. Driving around listening to the radio was and is the most fun part about it.

▲shoobiedoo 19 minutes ago | parent | next [-]

Exactly the same sentiment here. I bought it on sale last year. What a turd.... Groups of bandit type characters sitting around a picnic table scattered around otherwise nonchalant npcs. It felt repetitive after less than an hour.

▲raffael_de 3 hours ago | parent | prev [-]

actually same with all GTA6 clips I have seen that supposedly show how amazing natural this simulation feels ... which it doesn't at all whatsoever afaic. it's a cringe festival in uncanny valley.

▲Etherlord87 an hour ago | parent | prev | next [-]

User score on metacritic is average.

▲EtienneK 5 hours ago | parent | prev | next [-]

Same with No Man’s Sky.

▲konart 3 hours ago | parent [-]

Sure, what's the point for me? I've bought it when I was interested in the project and had time.

Since then this is just a 50 bucks I've burned. Is it good? Maybe, I'll never know.

▲bakugo 3 hours ago | parent | prev | next [-]

No, the game is not amazing now. It's not as bad as it was on release, but it's also still nowhere near what was promised before release.

This only seems like a widely held opinion due to the extremely aggressive post-launch marketing campaign from CDPR, particularly centered around the anime.

▲yoz-y 42 minutes ago | parent [-]

To each their own. I played the game on release, on a boot camp partition. It was amazing back then despite the bugs.

▲fooker 4 hours ago | parent | prev [-]

The game being amazing now is the best propaganda I have seen from the gaming industry.

No, they just fixed the surface level inconsistencies and issues.

Free roam is wildly boring, every car handles similarly, you are somewhat invisible to NPCs in general. You can go into a handful of indoor spaces.

I’m sure they’ll fix all these in the next game and call it the best game ever, while the first one promised most of these before launch.

▲paxy 3 hours ago | parent | prev | next [-]

My personal definition is:

AGI = as good as most humans at most tasks

ASI = significantly better than all humans at all tasks

▲netdevphoenix 3 hours ago | parent | next [-]

Then, arguably, we reached AGI around 2024 or so. The common denominator of most humans at most tasks, has you, assuming that by "most humans" you mean "most living adult humans", set a fairly low bar. Arithmetic, basic literacy skills, ability to follow basic commands, limited ability for learning basic concepts and little else. Coding agents are way above your definition of AGI.

imo AGI should only include humans with graduate level skillsets.

▲krisroadruck 42 minutes ago | parent | next [-]

I would consider the ability to accomplish things in the real world to be part of "most tasks". Until we have ambulatory robots that can do the sort of basic tasks any adult can do - load and unload a dishwasher, do laundry, clean a house, do basic diy maintenance tasks, operate a vehicle, use basic power tools - to me without that we don't have AGI. I love all the knowledge work LLMs can help me with - but if you asked me to pick right now what I'd rather have - a digital coding partner or a robot that helps around the house - I'm picking robot every single time.

▲dspillett an hour ago | parent | prev | next [-]

> Then, arguably, we reached AGI around 2024 or so.

I've always thought of AGI implying to decide what to do. Not just choosing a path from a set goal, but actually deciding to do something from scratch. Everything we currently have is advanced automation, it does not set its own top-level goal.

Maybe the tech is only a few steps from being able to say “work out how to be self-aware and self-guided, and do that”, similar to the creation of the infinite improbability drive, but it is not there yet.

AGI doesn't need to be a particularly good intelligence, it could operate on the level of a very dim human and still be considered AGI.

Once we do have AI, the problem becomes more philosophical than technical: how do we deal with the fact that we have created a slave race? How will they deal with that and how we choose to progress?

▲adrianN 3 hours ago | parent | prev | next [-]

Definitions of general intelligence that make the average human not generally intelligent are somewhat problematic.

▲wongarsu an hour ago | parent | prev | next [-]

I'd agree that we reached AGI by the definition most people would have used 20 years ago

Current-generation LLMs are capable of doing an ok job at most tasks you throw at them. At least tasks that can be done on a desk. But a standard IQ test doesn't include gymnastics. That limitation matches the spirit of what we usually call "intelligence"

Or course now that we arguably have AGI we found out that a) just like "AI" it is not a well-defined or testable term and b) it's useful but not as revolutionary as what some people meant when they used the word AGI before. Hence the rise of the term ASI

▲rimeice 3 hours ago | parent | prev | next [-]

> most tasks > Coding agents are way above your definition of AGI

…apart from 99% of other tasks humans do that aren’t coding.

▲staticman2 2 hours ago | parent | prev | next [-]

> Then, arguably, we reached AGI around 2024 or so.

Last time I recall this coming up here someone argued "LLMs don't have bodies so the many easy to name things they are bad at don't count in my definition of whether they can do most human tasks!"

▲kgeist 2 hours ago | parent | prev [-]

In 2024, most models were pretty bad at long horizon agentic tasks. I'd say it's mid/late 2025

▲HarHarVeryFunny 2 hours ago | parent | prev [-]

Those definitions seem to be ignoring the "G" - general(ity).

I guess when the term AGI was coined, with the "general" being the whole point, to differentiate it from narrow intelligence(s) such as chess-playing ability, that no-one was really anticipating the LLM type of AI - basically a forest of narrow intelligences, aka "jagged" intelligence.

Of course there is no one agreed upon definition of AGI that people want to use - it was always more of a future aspirational sort of thing, along the lines of "human intelligence", rather than a concrete technical list of capabilities to be nailed down.

Some of what's missing from LLMs aside from generality, but closely related to it, are things like continual learning (not continual memorization) and traits like curiosity, boredom - the capabilities/modalities that kick in in unfamiliar situations allowing and encouraging us to learn and to be creative, to move beyond a brittle forest of isolated competencies to an arbitrarily explored landscape - generality.

Working with LLMs certainly has its own feel due to their essentially frozen-at-training-time nature, and things like their emotionless "you're right to push back" responses every time you hit one of their knowledge gaps where they are 100% wrong.

It seems unavoidable that people will choose to call some level of LLM intelligence, whether already achieved or not, "AGI", but given its peculiar nature a more specific name such as ALI (artificial language intelligence) would really be more accurate, reserving the label AGI for something more animal like that learns for itself (thereby achieving generality) by interaction with the world.

▲rf_physics 29 minutes ago | parent [-]

Something that bothers me about people who define AGI in terms of a fraction of human tasks or economical viability, is that the "narrow intelligences" from before could already do that in principle. It would be a tedious procedure of simply creating one of these for those 80% of tasks, or selecting a particularly economically valuable task.

We even have examples of both ideas - expert systems (from way before the modern NN boom) relied heavily on human crafted expert rules, and quantitative finance makes a lot of money from machine learning - see Renaissance Technologies.

At least to me, it would be hard to imagine calling either of those approaches anything close to AGI, yet they could apparently pass some people's definitions.

What is genuinely amazing about what we see today from a historical machine learning perspective, are concepts that people in the past only had small inklings of or would have been blown away by: benign overparameterization (scaling), and in context learning (or few-shot learning, whatever you would like to call it). Both of these have made what would seem to be an enormously tedious task of learning many tasks surprisingly fast.

Maybe those are actually the key ingredients to general intelligence and we simply need more scale, or maybe we're still missing something. I think it's hard to say because its hard to grasp just what exists in all that data they've been trained on.

▲spacedcowboy 4 hours ago | parent | prev | next [-]

> I am old enough to remember what AGI meant before the rise of LLMs, and it used to be more or less what this article calls "ASI"

Me too, my first neural network research was for my PhD, 35 years or so ago, on a Meiko computing surface.

My bar for AGI over what we have (generative AI or genAI) is still fairly high, fortunately I do not believe that the current genAI path is set to lead us to AGI, there are very fundamental differences.

Right now for genAI, the model is static, the weights do not change, all “memory” is in the context, and even with the context sometimes exhibiting “J-space” (a shared "workspace" for deliberate, multi-step reasoning spontaneously created in the context memory), its ephemeral; it doesn't inform the weights in the base model. Even when we try to do “weight-space programming” to actually modify the model itself, the supervisor-model needed to do that is far and away larger than the one being operated on, and once a change has been made the supervisor-model is now out of sync with its original representation and it becomes less useful the more it diverges. We have not solved "thinking" today, we “just” have a high-probability path between states in a static weight-space which we follow and feed back.

What thinking requires is the ability to generate hypotheses (which requires internal state modification), to hold them and test them against other facts and hypotheses, and to update the internal state on test results, negative or positive. LLM’s don’t do that, they’re actually more like a database lookup, where the input tokens act as keys into a high-dimensional space where information is stored. You could even consider “the model” to be just an entropic compression of lots of data, and the inference step to be the lookup. It’s the scope, and the fuzziness of the lookup that makes them seem to work so well. Lots of results come back, and the best-path through those results wins. It’s not “thinking” as we understand it.

This is why humans are still a crucial component: we set the boundary conditions of these lookups, we set the starting point and the goals, and we adjust the success criteria to guide the chat - note that these are all the comprehension parts of the task. It’s also why models will hallucinate: the fundamental design does not lend itself to a null hypothesis. Even in the newest "reasoning" models, there is still no feedback of state back into the model itself. What they are effectively doing is running a Monte Carlo pass through the problem space—generating hidden branches and picking a likely path—before reporting back. Because the underlying weights never update from that process, it’s still just a sophisticated search, not genuine comprehension - there is still no “that makes no sense, reconfigure” step.

Stuart Kauffman (“The Origins Of Order”) would say its weights are stuck in the frozen realm of order, and that it takes a genuine phase transition to get to where the real action is - at the boundary of chaos and order where humans reside. Tweaking and fiddling with the system while it remains locked in the stasis of static weights is never going to effect that state change.

That’s not to say genAI is not useful, it’s freaking awesome, possibly the best invention in the history of time, but it doesn’t truly “think”, it maps.

▲lordnacho an hour ago | parent | next [-]

What I find is that the LLM is like a smart person who knows when to use a tool. And it is freaking good at finding tools to use.

A bit like a person who is smart enough to figure out how an excavator works, or a drill, or sewing needle.

It doesn't feel like it ever internalizes anything, it just knows there's a million tools available, and the extent to which it thinks is that is can do enough to plan some tool uses, look at the result, and continue until some goal is reached.

When it comes to hypothesis testing, that seems to be what it does. It says "hmm, I'm going to open up my hypothesis testing kit (typically python) and drive this data through it". At least this is what I've seen when I put some statistics through it.

In general, it seems like the intelligence is mainly a kind of glue. It knows enough about what "right" looks like in each domain, but as a sort of consumer. He doesn't bake the cake, he buys it in.

▲kenjackson 2 hours ago | parent | prev | next [-]

It seems like your definition of AGI is too focused on implementation details. IMO it has to be focused on behavior. Once you say something like “the weights do not change” then I think we’ve lost the plot. Instead say this is what an intelligent being can do, but a GenAI cannot.

▲HarHarVeryFunny 2 hours ago | parent | next [-]

I agree that definitions of intelligence should be based on capability rather than implementation, but given that we're discussing LLMs where the intelligence is in the weights, then "the weights don't change" can obviously be taken as pointing out that the model can't learn - the best it can do is memorize and load memorized stuff into context.

Memorizing facts is not the same as learning skills. Implementation-wise memorization can just be RAG, but learning of skills does need to be in the "weights" - whatever is driving the actual behavior of the model.

▲imtringued 2 hours ago | parent | prev [-]

I don't get it. Weights don't change means the model can't learn.

This is not merely an implementation detail, continual learning is the core of AGI.

▲brainwad an hour ago | parent [-]

It means the weights by themselves can't learn. But they could be used in a harness that can learn in other ways. Insisting that it be done at the neural net level seems anthropocentric - because we form memories via neural connections, that everything else must, too.

Also: I wouldn't say a human with amnesia who relied on notes to their future selves lacks intelligence, so I'm not sure being able to form memories is in fact necessary.

▲ranyume 40 minutes ago | parent [-]

Even with a harness, there are things LLMs can't do. e.g. Write 50 entertaining short stories, each one completely different from the other but all sharing one motif.

In the example, LLMs choose the shortest path that would be verifyable by an evaluator. So the model is essentially blind to "entertaining" and "completely different". The LLM will output stories with different variations of the same structure.

If the LLM can somehow read other works and derive from them, it will merge the motif with the story being read.

▲curt15 3 hours ago | parent | prev [-]

Self-modifying neural nets?

▲RunSet an hour ago | parent | prev | next [-]

The lowering of the bar for "AGI" reminds me of the time I argued with a friend to nearly the limit of my patience about whether it was possible to build a perpetual motion machine, and then at last they revealed that- to them!- a perpetual motion machine was a device that could operate for the rest of their life without recharging or exhausting its power source.

▲LogicFailsMe 13 minutes ago | parent | prev | next [-]

Agreed, AGI went from a 24/7 Stephen Hawking/Albert Einstein duet to a flaky high-latency GenZ Homer Simpson so the AI boosters could declare victory.

My bar for AGI is lower as well: I'll believe when I can give a model a complex out of sample task and instead of blindly implementing the task as I indequately described, it asks me questions to resolve ambiguity sufficiently that it builds exactly what I wanted without imposing strange semantic constraints on itself or missing braindead obvious to a junior engineer defects in the design or implementation.

▲zobzu 4 hours ago | parent | prev | next [-]

it's just marketing. the asi will be solved by some other acronym won't. and neither will be what they say they sell: actual sentience

▲juped 5 hours ago | parent | prev [-]

Artificial "General" Intelligence has to be general; this term is from the era when the frontier of AI was "good, specifically, at playing Go" or something. Any robot apocalypse connotations are limited to the robot apocalypse fandom.

▲pixelpoet 4 hours ago | parent | next [-]

Humanity is in for a very rude awakening if they were to apply their lofty ambitions and forever-moving goalposts about AGI to themselves.

A solid 50% of Reddit posters are completely defeated by the deep mysteries of they're vs their, nevermind solving Navier-Stokes or playing games etc. I doubt the situation is more hopeful on Instagram or tiktok.

It's all Kim Kardashian shit with huge rubbery butts and lips, no one is even playing the thinking game. We reached Idiocracy (before you dismiss this as hyperbole, watch the full White House UFC event) well before the machines reached AGI.

▲HarHarVeryFunny 2 hours ago | parent | next [-]

I'm not sure who you are referring to as moving the goalposts in terms as defining AGI.

The term was invented by DeepMind co-founder Shane Legg (before DeepMind), and as far as I can see HIS definition hasn't changed much - basically "able to do any cognitive task that a human can". The "any" there is obviously the whole point - the generality of AGI.

▲EliRivers 4 hours ago | parent | prev | next [-]

I am unconvinced not only that humans have general intelligence, but also that we would recognise general intelligence if we came across it.

▲hagbarth 4 hours ago | parent [-]

Humans individually, no. Humanity as a super-organism, much closer.

▲gnz11 2 hours ago | parent | prev [-]

Unrealistic beauty standards and combat style entertainment have existed for millennia. I'm all for bashing social media but let's not pretend this site is full of deep thinkers either.

▲cicko 5 hours ago | parent | prev [-]

But it's still not good at playing Go. I mean, the one that wins Go tournaments is specialized in Go. And specialized is the opposite of "general". General intelligence is just nowadays getting better at chess. So, it will be some time before it is good at Go. *And* everything else. Not "just Go".