I have a linguistics background and a lot of my philosophizing lately has been on whether or not the emergent abilities of the LLMs is deep down a similar mechanism that creates our consciousness.
For a little bit I was working on having linguistics based evals for a kaggle competition. My challenge was whether or not I could mask things well enough to not trigger its internal state of certain phenomena, and that sent me down a rabbit hole that I'm still exploring.
This story resonated with a lot of questions that can come out of figuring a good solid answer to the what is consciousness question. The one I triggered for me is: Is our perception of time just a slow thread in the giant GPU we are running the universe on? Or more generally, what is time? That's a fun YouTube rabbit hole if you ever need one.
https://www.edge.org/3rd_culture/ramachandran07/ramachandran...
In short, as far as I can remember: evolutionary, it makes sense to understand other humans, to feel what they feel(empathy - the mirror neurons system), and simulate their thinking and feelings.
And once we have those systems, we can also use those on ourselves. And that's consciousness.
Edit:And I wonder if this is a testable hypothesis. Let's say we create a virtual humans. Some will have only the capability to simulate other humans. Some will have that, plus a slight capability to use that system to simulate themselves. Other could use that system in more complex and full ways.
And let them be in a competitive evolutionary environment and measure their fitness
It's terrific, but the poetry is from the original it links to, in case you didn't realise.
It's a brilliant and timely update though.
Aside, there are various recorded versions including video on YouTube but this is my favourite, a radio play:
They're Made Out of Meat
https://www.wnycstudios.org/podcasts/studio/segments/168264-...
The self-modeling, is in such a tight loop, it melds "ourselves" and our model of ourselves, our thinking and choices, and experience of our thinking and choices, into one component.
Like you can't analyze half a wheel of a bicycle and be talking about the same thing.
This awareness, increased modeling, control, feedback loop has tightened up over many stages. Just a few:
1. The body-sense loop
2. The internalized-environment-model loop
3. The body-internal-function loop
4. The body-internal-model loop
5. The emotional-cognitive loop
6. And finally, the tightest loop of all, our high-level cognitive activity, experienced as feedback directly, our self-model, and our self-direction, all merged into one thing.
We literally spend almost all day, every day, thinking about ourselves, in terms of our inner self.
That is consciousness. Rich self awareness, a merger of self-model and self-direction, and all in service of understanding and managing ourselves. Hw we can leverage our greatest tool, our self-directable mind, its habits, views, and behavior.
This wasn't an accident. A happy side-effect of our brains. It is a biologically evolved focusing of our highest-level behavior, with tight feedback, constant self-modeling and continuous focus on our inner status as motivation and most privileged object of our control. It has been ruthlessly optimized for, for a very long time.
You would really like Michael Pollan's latest book [1], entirely devoted to his exploration of consciousness researchers' POVs on this exact topic.
My favorite quote is that ~"perhaps Descartes was only half-wrong when suggesting I think, therefore I am; it seems rather closer to I FEEL, therefore I am."~
[1] <https://www.amazon.com/World-Appears-Journey-into-Consciousn...>
----
I've grown thousands of plants; I've read two of the author's other books devoted to plants; in this book Pollan makes compelling arguments for plant sentience (over a much-longer timeframe).
Sure, perhaps plant consciousness is a bit of a stretch, but they're certainly intelligent and curious creatures. He makes both arguments supporting plant volition.
----
If you haven't seen My Octopus Teacher (Netflix), do. I'm a bald 275lb bluecollarguy... and I wept/awed (both). So beautiful, we bundled neurons.
----
Bonus quote ~"color is where reality and magic appear as-if together"~ [color isn't real, but is perceived]. We most-often see what's most-predictable, not necessarily what we actually detected [in the case of color: nothing but nanometers].
If consciousness really evolved gradually, you would expect to see for example dogs or gorillas having less of it, but if they has less of it, why does it function the same way? Like for example animals can be scared, happy, anxious etc, they can experience the full range of emotions and thoughts, so their conscious experience seems just as rich as ours. What I mean by this is, if you can be "less conscious", then what does that mean _exactly_? Is it that you have less content in consciousness, or is it that you feel more like you are asleep? Or something else? We don't have any examples in animals of "less conscious", I would argue.
This makes me think that rather than having emerged gradually, evolution found a mechanism by which consciousness exists, and then some animals have that mechanism and others don't. I think that if it is a mechanism, then this mechanism is located in one part of the brain, not many parts functioning together (though one possibility is that this mechanism coordinates brain activity in such a way to enable consciousness).
What we can do is simulate very simple brains by simulating relatively few neurons as they appear in worms. In this sense we are multiple magnitudes away where the increasing complexity implies exponential increasing difficulty.
I would think we are so far away that there will be unknown unknowns we encounter on the way.
https://www.quantamagazine.org/ai-is-nothing-like-a-brain-an... https://pmc.ncbi.nlm.nih.gov/articles/PMC9665914/
This is why making more neuromorphic NNs is still an active area of research, although they typically all focus on another extremely simplified model (spiking neural networks).
https://en.wikipedia.org/wiki/List_of_animals_by_number_of_n...
I was about to post the exact opposite question? How could it not be an emergent property? Unlike consciousness, the concept of emergence is pretty well defined: An emergent property is a characteristic or behavior that a complex system has, but which its individual components do not have on their own.
Consciousness itself doesn't have a well agreed upon definition, but I would posit that _most_ people would agree humans have it, and _most_ people would agree individual cells (neurons) do not have it. If you agree with those two statements, then consciousness is an emergent property by the definition I gave above
We can all agree on what color something is, but we can’t describe the color a priori, only by example. I think consciousness may be a similar phenomenon and the only test is by shared experience. If so then we are in deep trouble because we will not be able to anticipate when a system becomes conscious.
Why cannot this be applied to consciousness as well? I mean, it's surely much more difficult to do compared to colors but... impossible?
Consciousness can be not-emergent but also not metaphysical, think sci-fi-type undiscovered physics or matter.
Of course both of those suffer from the recursive problem of just kicking the can one level up. But I guess that's fundamentally unsolvable so who cares.
But I think my issue with the emergence theory is that it seems to imply to me that consciousness is non-physical and non-local. So what entity is actually experiencing the consciousness? It's not that I believe consciousness is physical and local, but people who make the emergence argument seem to believe it is and I can't figure out how that is supposed to work.
Or something like that. This gets to the "dorm room bullshitting" level right quick.
Idk, it's really hard to articulate my thoughts here and yes it is pretty close to the conversations I had in college on various substances. Lol.
And we are only doing it for a few decades. Evolution had million of years of "try and error".
Psychological time is your own weights being updated in response to stimuli and internal processing.
When there isn't anything interesting happening, no updates are needed, and you don't perceive much time. That's why there's a logarithmic effect on the "density" of time as you age.
Although part of me thinks some of this is from being substantially busier than ever (work + kids), and hoping maybe it can slow down again, at least a little bit.
Yeah, the weights not updating online makes them less like a living organism that can update and learn and evolve ... ok ....
What is no longer conscious, the brain? Or the body? Or some other entity?
If consciousness is weakly emergent, how do we know it emerges from the solely from the brain and not, say, brain + body? Or brain + body + or environment. Or from the universe itself?
I understand the math pretty well but still find it crazy that a bunch of matrices can converse in human languages without ever being “taught”.
Imagine decoding an encyclopedia written in a foreign language where the characters, punctuation, and grammar are unknown — supplemented by a million other texts the same way. Feels like it should be utterly impossible with any amount of computing power…
Today I asked my employer’s Claude to proofread a short software user manual written in markdown. (Trying this with a LLM was a first for me!) It pointed out not only grammar mistakes but also cases where I did not follow my own self-imposed conventions that were never explicitly stated. (I didn’t have a chapter detailing all the typographical conventions the way specification documents often do)
I also asked it what parts might be unclear to a user. The response was surprisingly good — no worse than asking the QA tester for the same feedback.
Also find the LLM seems to “comprehend” subtle technical details of obscure technical specification documents that nobody on the Internet ever discusses.
As for time and the universe, Stephen Wolfram’s theories seem intriguing. He seems a bit obsessed with pretty diagrams but the idea of time dilation being the result of computation seems somewhat more appealing than trying to imagine relationships between time, gravity, and the speed of light .
Proofread has a spot in that space, and layers allow patterns like terminology consistency to be expressed so your query will now tap into a subspace that will infer tokens based on whatever consistency patterns were ingested with proofreading texts.
There is a dictionary, it's called the tokenizer.
There are grammar rules, they are just very weak because the structure of human language is generally quite weak. When presented with languages which have strong consistent grammars the weights are very easily interpretable as a grammar: https://arxiv.org/abs/2201.02177
The point of the original short story is that the computational substrate doesn't matter when you have Turing completeness. This one seems to think that you don't need structure and interpretability just because you change substrates.
At best, it's a wordlist. It gives the LLM some idea of what humans consider to be common words. But it doesn't tell the LLM anything at all about those words. And it's not even comprehensive, many words map to multiple tokens. Nor is it exclusively words, some of those tokens are punctuation, or modifiers, or control tokens. On multimodal LLMs, some of the tokens actually represent image and audio data.
The LLM doesn't get informed about any of this up front, it has to learn what every single token means from context.
You are technically right, that it's something in an LLM that's not weights; But it's not that structured. And really it's only there so the LLM can interact with the outside world.
> There are grammar rules
There is no dedicated "grammar rule" structure in the LLM or the tokeniser. It has to learn them all from context, they get encoded as part of the 80 layers of weights.
That is your takeaway from the 1991 story?
fractally or factually? You mean wrong on so many levels you need a fractal to capture them? If so, what if you could use a neural network instead?
That paper did not train the models on 'a language with strong consistent grammars'. Mathematical Operation tables are not a language. Grammar itself is a post-hoc rationalization and there's no evidence LLMs follow 'grammar rules' anymore than the brain follows grammar rules. Of Course, that's not to say transformers can't learn simple rules if the dataset calls for it.
Not a natural language, but they are certainly a language as in a symbolic representation of information.
The tokenizer is, at best, a sensory mechanism as evidenced by 1) the random generation of the tokenization scheme, and 2) vastly different tokenization schemes produce virtually identical behavior. It'd be like if Noah Webster threw a bunch of movable type into a bucket (breaking some words in half) and then drew randomly to make the first English dictionary.
EDIT; I was too cavalier with the comparison of tokenizer to sensory modality; my ultimate point is that direct byte-to-token transformers can achieve similar overall performance which to me makes a weights to meat comparison pretty straightforward, but the particular tokenizer in use certainly has a large impact on both efficiency and accuracy on specific problems (e.g. digit representation)
So when I way that the grok paper and the pong paper fundamentally agree I have some idea of what I'm talking about.
It's just that the rules we feed in the model are extremely poorly defined and we end up with the soup of disjoint rules smeared all across the weights.
This isn't a feature of the models. It's a feature of the training set.
Being shocked that you can store rules in floating point numbers is the same as being shocked you can store rules in integers. It's been a century since Goedel Numbering was invented, we should be used to it by now.
That statement caught my eye. It's either trivially true or quite clearly wrong, depending on how you mean it.
In the literal meaning it's true. Given any finite set of real numbers, I can easily produce a different set (like taking the original set and adding a number which wasn't in there like one plus the largest or so) from which you can trivially produce the original set computationally.
But if you mean you give me both sets then that can't be true. For example if you give me a single real number as set A and the empty set as set B then I can't create a program which generates set A from set B. Your real number in set A could encode anything.
It's a learned mapping from one representation to another, not some semantic lookup against an exogenous source.
Or to echo article, the dictionary is made out of weights.
You can't move your mind to and any other brain, but weights can run on any GPU.
And they're made out of weights.
The 'magic' in weights is that the rules are spread through the whole model and you can't point to one place which encodes them.
The grokking paper shows that this stops being the case with enough training data and enough compute.
https://web.mit.edu/people/dpolicar/writing/prose/text/think...
It stars Tom Noonan and Ben Bailey!
Because we are not taking things seriously. If ClosedAI or DeepDisTrust or Posthropic come up with something that quacks like a sentient being, our built-in innate reaction is going to be to scorn it, dismiss it and end the conversation. The alternative, to even consider that we fungible creatures who live in apple-eating-sin that got us expelled from Eden can create alien souls, souls that are at the very least our equals, would be teleological Armageddon. It would force us to acknowledge the mutable nature of souls and the malleability of being. We would have to stop believing that the nature of disease and death is more divine than ourselves.
Do those actually qualify as alien, if they're products of our human culture and just the substrate is different?
> We would have to stop believing that the nature of disease and death is more divine than ourselves.
Why? Stopping believing in mutually contradictory claims is not a requirement. Especially when it comes to concepts that don't seem to have a definition, like "divine".
But congrats: this is absolutely & incredibly brilliant.
Can't wait for the Jon Benjamin voiceover.
- Terry Bisson, 1991
https://web.mit.edu/people/dpolicar/writing/prose/text/think...
Radio play by Miriam Tolan and Russ Armstrong:
https://www.wnycstudios.org/podcasts/studio/segments/168264-...
(EDIT: the original parent was missing "rather adapting from the original")
Here is Jon Benjamin reading Bisson's original text: <https://www.youtube.com/watch?v=5usXhX0zaO4>
From a circuit perspective that makes kinda sense, but from the abstract "bit" perspective, the "switching bit" is a mechanism that operates on bits which in the end are also data. In other words there is only one type of bit: the data bit, and the switching comes on top of it.
Not really. What usually flows (in metals) are electrons. Quarks stay where they are. And when we prefer to think about flow of positive charges, the positive charge in question is a hole left by a missing electron. Physically real positive charges (ions) can flow in electrolytes though.
Bravo. Really helps (even with my own) perceptions of newness. Similar to stsitned short-story (on dentists, backwards).
Just tokens produced by weights.
Useful, but never forget that ground truth!
https://web.mit.edu/people/dpolicar/writing/prose/text/think...
> These models are the only other things we've ever met that can hold a conversation, and they're made out of weights
Is a fair point.
Parrots are intelligent animals, albeit with a limited capacity for vocabulary and syntax compared to a human, and Eliza and the flowchart are made out of explicitly encoded rules and conversational tactics.
The above take fails in the real world because neuronal cells don’t exist in a vacuum; they are products of cellular development from a zygotic union of haploid contributors of sequential genetic information optimized for survival in an oxygen-rich biosphere powered largely by our local star that supports mammalian life (and microbial, plant, avian, etc.). Real AI would thus be AL - artificial life - as much as artificial intelligence. I don’t think you can have the one without the other, which upsets the simulationists who think an agent in the Matrix would be intelligent.
What either interpretation implies is that any real ‘artificial’ intelligence would be no more artificial than you or I, but it would have to dynamically update its weights at the same speed a human nervous system could (think how quickly we learn not to poke a cactus). For it to be at all trustworthy, then like a human, it would have to undergo a socialization process, one of the results of which is the development of a sense of embarrassment when it breaks acceptable social norms.
Hmm, this reminds me of the recent statement of the Pope about AI, of which I immediately thought, “Wait a second, aren’t there a fair number of people like this? The narcissistic sociopath profile, I think it’s called, a bit unfair to assume any real AI would turn out this way, isn’t it?”
Pope: “ Nor do they have a moral conscience, since they do not judge good and evil, grasp the ultimate meaning of situations, or bear responsibility for consequences. They may imitate or even simulate, but they do not understand what they produce, for they lack the affective, relational, and spiritual perspective through which human beings grow in wisdom.”