Artificial-intelligence systems such as large language models (LLMs) can exhibit behaviour that seems strikingly intelligent. This has led many people, including some experts, to wonder whether these systems could have some form of consciousness — the capacity for subjective experience. However, we think that existing AI systems are not conscious, and that, along current trajectories of development, they are not likely to become conscious. Greater clarity about whether AI could become consciousness will be crucial as society grapples with the challenges posed by this increasingly powerful technology.
As others have pointed out we do not even agree on what consciousness is. There are three competing camps:
- Mechanists - if it can pass mechanical tests it is conscious
- Dualist - you need more than just the brain, some second soul-like thing to be conscious
- Pan psychist - everything is conscious
If you are in camp 1 the conversation about consciousness is getting more complicated as AI passes more tests. If you are in camp 2 AI can never be conscious. If you are in camp 3 AI and everything else is conscious including rocks and rivers and the stars in the sky.
I think arguments like this are misleading when discussing the question of AI possessing consciousness. Materialism, dualism, panpsychism, idealism, etc are all philosophical concepts; they attempt to answer “what is the nature of consciousness and how does it arise from the body?” from a philosophical standpoint. All of the various options have pros and cons and there is no clear “winner.”
However, there are decent scientific definitions for consciousness. These attempt to answer the question of “what do we mean when we say something is conscious?” from a scientific standpoint. There are criteria that neuroscientists, doctors, etc generally agree on.
So it’s not really correct to say “no one knows what consciousness is” when evaluating whether something is conscious. And I think it’s quite clear that current LLM algorithms could never be conscious, using the current accepted scientific criteria.
(I’m not a philosopher or a neuroscientist, but I do have a degree in computer science and have a solid understanding of how LLMs work.)
Ok, I’ll ask. What are these decent scientific definitions for consciousness and how can we use them to construct a test for consciousness?
You are free to search the web, there’s lots of information out there. You might start with the Wikipedia article on consciousness.
Oh, I thought you might have something new to say. If you’re just using the normal definition and not bringing any sort of testability (which imo is impossible) or new arguments, how exactly does the common definition prove anything must be non-conscious? That seems like the main claim of your comment, yet it seems like an unsupported assertion; if the common definition of consciousness could easily prove things not to be conscious as you seem to imply, this discussion would not be happening at all.
From Wikipedia:
The Cambridge English Dictionary defines consciousness as “the state of being awake, thinking, and knowing what is happening around you”, as well as “the state of understanding and realizing something”.
Most of these are trivially easy to test (not sure why you think that’s impossible). And LLMs have none of those properties.
Later:
“The having of perceptions, thoughts, and feelings; awareness.”
Again, LLMs have none of these things. They simply generate a probable natural-language response based on their training, deterministically.
Feel free to do further research on your own if you’re interested.
Feel free to do further research on your own if you’re interested.
Philosophy is one of my hobbies and consciousness one of my fascinations. Respectfully, doing “research” by skimming the wikipedia article would be a waste of our time; as would responding to any particular definition and arguing over the individual words people have used in summaries of the topic. So I’ll skip that phase of the convo.
The fundamental problem with measuring consciousness is that it’s an internal state (a “qualia”*). It is fundamentally an experience, something that exists only as a concept in your mind. This makes it immeasurable.
That’s quite the claim, so I’ll provide a smaller concrete example: color. Obviously you can measure the frequency of a photon and say “ah yes, 700nm, that photon is red”. But strictly speaking, the photon isn’t red, it just causes you to percieve “red”; it has a wavelength which triggers your eyes and brain to produce the subjective experience of the color red. That subjective experience of color is a qualia, and it is not directly measurable; at best you can measure the phenomenon that causes that qualia in you, and assume that that phenomenon would cause the same qualia in someone else when they percieve that photon. Technically, you can’t be sure that this assumption is true-- this leads to the “Inverted Spectrum” thought experiment (another great wiki page if you prefer their explanations to mine), where someone’s subjective experience of color as it relates to the frequency of light could be entirely different to your own.
Anyways, you cannot prove the existence of nonexistence of such an internal experience, and consciousness belongs to this category of qualia. You know that you are conscious because you experience being conscious, not because you can prove it by looking at what you do. You can’t even be sure that I’m conscious, even if you met me in real life; you can’t prove that I am actually thinking, feeling, understanding, knowing, or realizing anything, because you cannot be me. You can only assume that I am doing those things because you observe me doing stuff in physical reality that’s consistent with the way the only entity you can be sure is conscious (you) is acting, and assuming that since I’m made of the same stuff as you and I act kind of like you do that I am probably conscious.
If you could actually prove that it is impossible for a machine to be conscious without that same argument applying to humans, then the philosophy community of the last several millenia would praise you as their messiah. I am interested to see what your argument would be if you want to give it a shot though-- no harm in trying.
* Like “data”/“datum”, technically the singular form is “quale”, but I’m gonna be real, nobody ever says that. Anyways if you like reading wikipedia pages about philosophy, “qualia” is a good one!
How is red a subjective experience? We can all experience red things, at least, all of us with functioning eyes. What about that is subjective? How is color not directly measurable? Color is very easy to measure, we can even assign it values, like (r,g,b). If you see something red, you have already measured it to have the values (>0,0,0). I don’t know what the term “internal experience” even means. Experience is just something that happens to something else. A rock experiences erosion in the rain. I am experiencing tiredness right now. To say something experiences something is to just say X is happening to Y. I don’t see how it is meaningful to assign adjectives like “internal” or “external” to it. I mean, if the adjective was meaningful, then would it not imply that you are implicitly contrasting “internal experience” with “external experience”? What on earth would external experience be? If you cannot tell me what external experience is, then the adjective “internal” is contrasting it with nothing at all, so it is a useless adjective.
Nobody can even agree on a concrete definition of consciousness so I’m not going to bother reading these ‘reasons’.
LLMs only respond to inputs. They don’t have independent ideas. And their responses are just predictive of what the input intends a response to be.
Accurate prediction models can be extremely useful, but there’s a reason why weather prediction models aren’t considered intelligent. It’s because they aren’t designed to mimic human conversation.
They are playing us for absolute fools.
That first sentence describes all life. We only react to input. We don’t have independent ideas. We have the illusion of them, because we can’t see our own neural networks. Asked to explain your actions, you would give your best guesses, but you don’t actually know. You’re a biological, finite-state machine, moving from state-to-state based on stimulii, while supposing you are a thinking individual in control.
Citation needed on all of this, and even if you have citations, LLMs are deterministic, and do not learn or change over time. They can’t even be said to exist continuously since they only exist during the brief time that they’re responding to inputs, and then they disappear until the next input. I would argue that all of these things would need to change in order for them to be able to have consciousness.
They can learn and change over time, that’s just not enabled. The services that offer them want a reliable, reproducible, controllable experience. There’s zero reason they can’t continue to change through usage, other than it’s a conscious decision not to enable that functionality.
That’s why they can only be conscious when they train using new data, and run asynchronously with continual input - the same way we do.
Edit: People downvote me, and I suspect it’s because they’re thinking of the public facing models. I am talking about progress behind closed doors. Even some of the public facing models are gaining the ability to learn, via a mechanism being called memories. It’s just a short text note that is passed into the context window of every prompt, but isn’t that exactly what our short term memory is? In my view is no different. Pruning the inclusion list to include only the relevant information is the equivalent to use forgetting some information, but if it’s kept in long term storage with reference links to other memories, it can be “remembered” and brought back into the current context. It’s functionally the same as our own learning and thinking.
Even if we could update an LLM’s weights continuously, the algorithm is still deterministic, like all computer algorithms. There is no evidence that humans are deterministic (or a “finite state machine”) in any way. So I don’t think such a dynamically-updating LLM could be conscious, either.
It is my personal belief that all life is deterministic. It’s the only plausible explanation for instinctual behavior - like monarch butterflies returning to their breeding grounds they’ve never been to. Or people yawning because others yawn. These aren’t conscious decisions, they’re peeks at the deterministic coding that life operates upon.
I don’t think you understand what “deterministic” means.
A computer algorithm can absolutely be non deterministic, pick your language and run random.rand()
There is a whole class of such algorithms https://en.wikipedia.org/wiki/Nondeterministic_Turing_machine
But I don’t see how it makes a difference, even deterministic algorithms can produce infinite complexity, eg linear neural networks are proven to be able to compute any computable function, or the Mandelbrot set which is produced from a single recurrent statement but encodes a space of infinite complexity
Maybe your point is to say that a brain is performing a non computable function…but since your brain is computing that function…is that a counter example?
Reasons given:
- Intelligence is not consciousness
- Silicon brains are not made of meat
- Consciousness requires meat (and/or a particular winding of neurons)
These are, to be frank, really weak. They say on point 2:
Computational functionalism … is popular among some scientists and philosophers, too, but we think it is fatally flawed because brains are not — or at least not just — computers that are made of meat.
One main reason is that it is impossible to separate what biological brains do — their functional organization — from what they are — their material substrate. In other words, there is no distinct separation between the mind and living tissue in the human brain, as there is between software and hardware in a digital computer.
Restating clearly: Ghosts are real. The mitochondria is not only the powerhouse of the cell, it also gives you mystic powers.
Really? The source they site for this point features the lovely assumption “consciousness is unique to biological life”. From this axiom, I see how the conclusion follows, but I don’t see why we should believe this axiom. We can do physics emulation of every atom in a fruit fly on a computer…
This analysis is correct but I have a minor quibble: The simulated fruit fly brain is not a complete simulation of the biological processes, or even all the neuronal connections, of the actual fly’s brain.
@Feathercrown
@artISAn #consciousness #AI #brain #neuroscience #computer #functionalismEven a “complete simulation” of a fruit fly brain would be no more than an approximation. I think you mean “emulation” and I agree.
The simple reason is that we do not know scientifically all of necessary and sufficient functional mechanisms of consciousness, nor consciousness’s functional role in “intelligence” (which we also have trouble describing). If, and only if we we did, we would (given the right tools) be able to replicate it functionally (in principle) using computation.
Specifically, I refer to a Universal Turing Machine (UTM) and the Church-Turing Thesis (the foundational principle of all computer science). EVEN THEN, there are computational problems that cannot be proved to be solvable by a UTM (e.g., the Halting Problem) and we don’t know what, if any role they play in consciousness.
Bottom line for me: we are going to need a revolution in each of the sciences of psychoneurophysics, mathematical logic, and theory of computation before we can begin to talk about replicating consciousness in a machine.
I don’t see why we would necessarily need to understand consciousness to create it. There is no evidence that nature understands the consciousnesses it has created. In fact a human can create a new consciousness in about 9 months without knowing any theory.
The question in my mind is, how could we possibly tell we’ve succeeded? We can’t even say for sure that anyone other than ourselves is conscious.
@Feathercrown Maybe all you need are the right ingredients? I don’t think so. There must be a recipe. Human ontogenesis proceeds according to a process specified in our genes, modified a bit by the developmental environment. We don’t fully understand that process completely (AFIK) but there are no intrinsic obstacles. #Mind emerges at some point.
You are right about the question of recognizing consciousness. People are predisposed to attribute life, intentionality, and all kinds of “mental states” to all kinds of inanimate objects given the flimsiest evidence. The more something resembles ourselves, the quicker and deeper the belief is held.
I’m not sure consciousness is even required. It may be an illusion. It may be an artifact. It does not matter to the prospect of extremely powerful #AI systems. I fully expect models to assert consciousness and make demands on that basis. True or false, what do we do in that situation?
I’m not sure our society is ready to answer that sort of question.
@Feathercrown We are not ready, in the sense of thinking about the question thoroughly, but we don’t get to choose when it comes. I am hopeful, however, seeing the progress humanity has made in recognizing intelligence and self-awareness in other species. The human-centric bias is on the wane.
