[del82]

I mean, I think the reason I would say the night sky is “beautiful” is because the meaning of the word for me is constructed from the experiences I’ve had in which I’ve heard other people use the word. So I’d agree that the night sky is “beautiful”, but not because I somehow have access to a deeper meaning of the word or the sky than an LLM does.

As someone who (long ago) studied philosophy of mind and (Chomskian) linguistics, it’s striking how much LLMs have shrunk the space available to people who want to maintain that the brain is special & there’s a qualitative (rather than just quantitative) difference between mind and machine and yet still be monists.

[FloorEgg]

The more I learn about AI, biology and the brain, the more it seems to me that the difference between life and machines is just complexity.

People are just really really complex machines.

However there are clearly qualitative differences between the human mind and any machines we know of yet, and those qualitative differences are emergent properties, in the same way that a rabbit is qualitatively different than a stone or a chunk of wood.

I also think most of the recent AI experts/optimists underestimate how complex the mind is. I'm not at the cutting edge of how LLMs are being trained and architected, but the sense I have is we haven't modelled the diversity of connections in the mind or diversity of cell types. E.g. Transcriptomic diversity of cell types across the adult human brain (Siletti et al., 2023, Science)

[simonh]

I’d say sophistication.

Observing the landscape enables us to spot useful resources and terrain features, or spot dangers and predators. We are afraid of dark enclosed spaces because they could hide dangers. Our ancestors with appropriate responses were more likely to survive.

A huge limitation of LLMs is that they have no ability to dynamically engage with the world. We’re not just passive observers, we’re participants in our environment and we learn from testing that environment through action. I know there are experiments with AIs doing this, and in a sense game playing AIs are learning about model worlds through action in them.

[FloorEgg]

The idea I keep coming back to is that as far as we know it took roughly 100k-1M years for anatomically modern humans to evolve language, abstract thinking, information systems, etc. (equivalent to LLMs), but it took 100M-1B years to evolve from the first multi-celled organisms to anatomically modern humans.

In other words, human level embodiment (internal modelling of the real world and ability to navigate it) is likely at least 1000x harder than modelling human language and abstract knowledge.

And to build further on what you are saying, the way LLMs are trained and then used, they seem a bit more like DNA than the human brain in terms of how the "learning" is being done. An instance of an LLM is like a copy of DNA trained on a play of many generations of experience.

So it seems there are at least four things not yet worked out re AI reaching human level "AGI":

1) The number of weights (synapses) and parameters (neurons) needs to grow by orders of magnitude

2) We need new analogs that mimic the brains diversity of cell types and communication modes

3) We need to solve the embodiment problem, which is far from trivial and not fully understood

4) We need efficient ways for the system to continuously learn (an analog for neuroplasticity)

It may be that these are mutually reinforcing, in that solving #1 and #2 makes a lot of progress towards #3 and #4. I also suspect that #4 is economical, in that if the cost to train a GPT-5 level model was 1,000,000 cheaper, then maybe everyone could have one that's continuously learning (and diverging), rather than everyone sharing the same training run that's static once complete.

All of this to say I still consider LLMs "intelligent", just a different kind and less complex intelligence than humans.

[kla-s]

Id also add that 5) We need some sense of truth.

Im not quite sure if the current paradigm of LLMs are robust enough given the recent Anthropic Paper about the effect of data quality or rather the lack thereof, that a small bad sample can poison the well and that this doesn’t get better with more data. Especially in conjunction with 4) some sense of truth becomes crucial in my eyes (Question in my eyes is how does this work? Something verifiable and understandable like lean would be great but how does this work with more fuzzy topics…).

[FloorEgg]

That's a segue into an important and rich philosophical space...

What is truth? Can it be attained, or only approached?

Can truth be approached (progress made towards truth) without interacting with reality?

The only shared truth seeking algorithm I know is the scientific method, which breaks down truth into two categories (my words here):

1) truth about what happened (controlled documented experiments) And 2) truth about how reality works (predictive powers)

In contrast to something like Karl friston free energy principle, which is more of a single unit truth seeking (more like predictive capability seeking) model.

So it seems like truth isn't an input to AI so much as it's an output, and it can't be attained, only approached.

But maybe you don't mean truth so much as a capability to definitively prove, in which case I agree and I think that's worth adding. Somehow integrating formal theorem proving algorithms into the architecture would probably be part of what enables AI to dramatically exceed human capabilities.

[simonh]

I think that in some senses truth is associated with action in the world. That’s how we test our hypotheses. Not just in science, in terms of empirical adequacy, but even as children and adults. We learn from experience of doing, not just rote, and we associate effectiveness with truth. That’s not a perfect heuristic, but it’s better than just floating in a sea of propositions as current LLMs largely are.

[pbhjpbhj]

>A huge limitation of LLMs is that they have no ability to dynamically engage with the world.

They can ask for input, they can choose URLs to access and interpret results in both situations. Whilst very limited, that is engagement.

Think about someone with physical impairments, like Hawking (the now dead theoretical physicist) had. You could have similar impairments from birth and still, I conjecture, be analytically one of the greatest minds of a generation.

If you were locked in a room {a non-Chinese room!}, with your physical needs met, but could speak with anyone around the World, and of course use the internet, whilst you'd have limits to your enjoyment of life I don't think you'd be limited in the capabilities of your mind. You'd have limited understanding of social aspects to life (and physical aspects - touch, pain), but perhaps no more than some of us already do.

[skissane]

> A huge limitation of LLMs is that they have no ability to dynamically engage with the world.

A pure LLM is static and can’t learn, but give an agent a read-write data store and suddenly it can actually learn things-give it a markdown file of “learnings”, prompt it to consider updating the file at the end of each interaction, then load it into the context at the start of the next… (and that’s a really basic implementation of the idea, there are much more complex versions of the same thing)

[TheOtherHobbes]

That's going to run into context limitations fairly quickly. Even if you distill the knowledge.

True learning would mean constant dynamic training of the full system. That's essentially the difference between LLM training and human learning. LLM training is one-shot, human learning is continuous.

The other big difference is that human learning is embodied. We get physical experiences of everything in 3D + time, which means every human has embedded pre-rational models of gravity, momentum, rotation, heat, friction, and other basic physical concepts.

We also learn to associate relationship situations with the endocrine system changes we call emotions.

The ability to formalise those abstractions and manipulate them symbolically comes much later, if it happens at all. It's very much the plus pack for human experience and isn't part of the basic package.

LLMs start from the other end - from that one limited set of symbols we call written language.

It turns out a fair amount of experience is encoded in the structures of written language, so language training can abstract that. But language is the lossy ad hoc representation of the underlying experiences, and using symbol statistics exclusively is a dead end.

Multimodal training still isn't physical. 2D video models still glitch noticeably because they don't have a 3D world to refer to. The glitching will always be there until training becomes truly 3D.

[skissane]

An LLM agent could be given a tool for self-finetuning… it could construct a training dataset, use it to build a LORA/etc, and then use the LORA for inference… that’s getting closer to your ideal

[ako]

Yes, and give it tools and it can sense and interact with its surroundings.

[FloorEgg]

Oh, I just realized you maybe we're referring to Kopple when you said sophistication?

If so, then yes, that might be a good measure. I'm not deep enough in this to have an opinion on if it's the best measure. There are a few integrated information theories and I am still getting my head wrapped around them...

[subjectivationx]

I think the main mistake with this is that the concept of a "complex machine" has no meaning.

A “machine” is precisely what eliminates complexity by design. "People are complex machines" already has no meaning and then adding just and really doesn't make the statement more meaningful it makes it even more confused and meaningless.

The older I get the more obvious it becomes the idea of a "thinking machine" is a meaningless absurdity.

What we really think we want is a type of synthetic biological thinking organism that somehow still inherits the useful properties of a machine. If we say it that way though the absurdity is obvious and no one alive reading this will ever witness anything like that. Then we wouldn't be able to pretend we live at some special time in history that gets to see the birth of this new organism.

[FloorEgg]

I think we are talking past each other a bit, probably because we have been exposed to different sets of information on a very complicated and diverse topic.

Have you ever explored the visual simulations of what goes on inside a cell or in protein interactions?

For example what happens inside a cell leading up to mitosis?

https://m.youtube.com/user/RCSBProteinDataBank

Is a pretty cool resource, I recommend the shorter videos of the visual simulations.

This category of perspective is critical to the point I was making. Another might be the meaning / definition of complexity, which I don't think is well understood yet and might be the crux. For me to say "the difference between life and what we call machines is just complexity" would require the same understanding of "complexity" to have shared meaning.

I'm not exactly sure what complexity is, and I'm not sure anyone does yet, but the closest I feel I've come is maybe integrated information theory, and some loose concept of functional information density.

So while it probably seemed like I was making a shallow case at a surface level, I was actually trying to convey that when one digs into science at all levels of abstraction, the differences between life and machines seem to fall more on a spectrum.

[foogazi]

> I think the reason I would say the night sky is “beautiful” is because the meaning of the word for me is constructed from the experiences I’ve had in which I’ve heard other people use the word.

Ok but you don’t look at every night sky or every sunset and say “wow that’s beautiful”

There’s a quality to it - not because you heard someone say it but because you experience it

[TeMPOraL]

> Ok but you don’t look at every night sky or every sunset and say “wow that’s beautiful

Exactly - because it's a semantic shorthand. Sunsets are fucking boring, ugly, transient phenomena. Watching a sunset while feeling safe and relaxed, maybe in a company of your love interest who's just as high on endorphins as you are right now - this is what feels beautiful. This is a sunset that's beautiful. But the sunset is just a pointer to the experience, something others can relate to, not actually the source of it.

[drewbeck]

I’ve seen incredible sunsets while stressed depressed and worse. Are you saying sunsets cannot be experienced as beautiful on their own?

[adastra22]

Because words are much lower bandwidth than speech. But if you were “told” about a sunset by means of a Matrix style direct mind uploading of an experience, it would seem just as real and vivid. That’s a quantitative difference in bandwidth, not a qualitative difference in character.

[holler]

my thought exactly

[dmkii]

It’s interesting you mention linguistics because I feel a lot of the discussions around AI come back to early 20th century linguistics debates between Russel, Wittgenstein and later Chomsky. I tend to side with (later) Wittgenstein’s perception that language is inherently a social construct. He gives the example of a “game” where there’s no meaningful overlap between e.g. Olympic Games and Monopoly, yet we understand very well what game we’re talking about because of our social constructs. I would argue that LLMs are highly effective at understanding (or at least emulating) social constructs because of their training data. That makes them excellent at language even without a full understanding of the world.

[heyjamesknight]

You don’t have a deeper “meaning of the word,” you have an actual experience of beauty. Three word is just a label for the thing you, me, and other humans have experienced.

The machine has no experience.

[intended]

The fact that things are constructed by neurons in the brain, and are a representation of other things - does not preclude your representation from being deeper and richer than LLM representations.

The patterns in experience are reduced to some dimensions in an LLM (or generative model). They do not capture all the dimensions - because the representation itself is a capture of another representation.

Personally, I have no need to reassure myself whether I am a special snowflake or not.

Whatever snowflake I am, I strongly prefer accuracy in my analogies of technology. GenAI does not capture a model of the world, it captures a model of the training data.

If video tools were that good, they would have started with voxels.