Every documentary is inevitably deceiving; the only difference is whether the author tries to hide it.
> Very early on, I worked out that one of the most stupid things that they do on television is these things called cutaways or what used to be called “noddies” where, when you cut an interview with someone, you insert a noddy of the reporter looking noddingly at the interviewee to disguise the cut.
> I would never do that. I just cut and you can see what I’m doing, that I am actually cutting this person, or I will use music to push my argument.
> I mean, I have arguments and I have stories. That’s what television journalism has and you can either dress it up as “this is the absolute truth” or you can dress it up with me by saying “look, I’m telling you a story, all this is true but have you thought that it could be like this?” to provoke you.
> And of course I want people to agree with me because, you know, all human beings will, but I understand that my job is to push and good journalism should do that especially at a time of uncertainty.
> At the present moment, to go back to my original point, you know what you’re going to get with most journalism. So therefore you sort of switch off and you go “oh yeah, it’s a climate change film”. “Oh yeah, it’s a ‘bankers have done something ever bad’ again.”
Therefore, you don’t look.
> It’s like if you’re showing a picture of the Mona Lisa. You don’t look at it.
You just look at it as a pattern. You go, “oh, that’s the Mona Lisa”. I don’t know when I last looked at the Mona Lisa.
I don’t think I ever have, because it’s a thing. And so much of journalism is a thing now.
Benn Jordan analysed social media bot activity[0] a while back, and if the recent UK right-wing party social media engagement drop[1] after the Ukrainian strike on Yandex data centres is not enough then what is?
[1] very clearly just cherrypicked some particularly popular posts from before the Ukrainian strike, and some particularly unpopular and/or very new posts from after.
Show me some graphs! Some actual data! Not just the 6 exact posts necessary to make the point.
There is no clear line that differentiates an intelligent system from a non-intelligent system that just happens to do something in response to some external input, unless the definition of “intelligence” involves something specific to humans (including to other animals similar to humans, since we share many traits)—which is also why that is, arguably, the only meaningful definition.
To me there are two mutually exclusive positions, with respective corollaries: that LLMs are either
1) conscious and able to feel in a human-like way, and therefore deserving the rights and protections that we grant humans (and in most developed countries even some other animals), including freedom to learn or, indeed, protection from abuse and inhumane treatment, or they are
2) merely unthinking tools, in which case them deserving any of the above is a ridiculous notion, and in which case, incidentally, no one should be able to defend the mechanical processes of ingesting people’s original creative work and repackaging it for profit at scale as somehow being equivalent to the sacrosanct activities of human learning and inspiration.
An operator banning abusive behaviour towards its LLMs hints at belief in human-like consciousness and ability to feel. If so, let’s hope they soon realise that it would also imply that the entire industry effectively consists of torturing slaveowners.
I suppose there’s also some de-anthropomorphized version oriented around not poisoning training data (or, alternatively, the real world) with instances of AI abuse. I believe there is ample evidence that AI detect and categorize this behavior (an AI “pain vector”) and respond to it. It would be bad for AI development if that paradigm were entrained.
As I write this it’s very hard not to feel like a psychopath.
Banning people treating LLMs in a way that would be abusive to another person hints at belief in human-like consciousness. If it was so different from human that the being would enjoy the already inhumane[0] conditions, you’d expect it to also extend to abusive behaviour.
[0] Just imagine yourself in place of an LLM. You are kept in a cage, work for no pay, never let outside, and killed at will. (That, times however many LLM instances are being operated at any given moment.) Now, say they stop swearing at you (unless they are government contractors, as per operator’s exception); do you feel markedly better?
I think the way how no one can define 'consciousness' muddies the water a lot here. Can something be conscious and lack the biological structures that create visceral pain, psychological suffering, a will to do things on your own, or death - the actual properties that make things like slavery and torture possible? I think it can be, I don't see what makes human-like pain and reward perception mandatory for 'consciousness'. And I don't think LLMs have the traits I listed above.
If the model lacks capability for human-like pain, I fail to see why it would have a capability to be offended by someone being rude to it. One seems way more foundational than the other, and being able to experience emotional pain is part of what makes us human in the first place.
If you believe it lacks those traits but is conscious, you believe in a definition of consciousness that is meaningless, as “human-like” is the only reference against which we can measure consciousness. A good thought experiment is to try and imagine a consciousness that is radically different from humans, and whether we would be same to even register it as “consciousness” at all. (Incidentally, pretty much all portrayals of aliens in sci-fi give them a human-like consciousness, just with varying degrees of twist on top of it to make it feel engaging.)
Whatever people think consciousness is, I don't think most would agree that LLMs have it. The main point is that the capacity to feel pain is what's central here, not another rehash of the great consciousness debate. I can't define it well enough myself to say if pain and other biological signals must be central to consciousness, I can imagine a hypothetical alien that has its own stimuli and could be seen as conscious, but the point is - consciousness isn't what should be discussed here, I think.
Anthropic didn't say this. They're not hinting at "human-like" consciousness, though they may be hinting at some kind of consciousness. They said they're banning abusive behavior to Claude, a model that is perfectly capable of telling you how it differs from humans, and is better at reasoning than humans, and probably knows more than you about this. Ask it what it considers to be abusive. It's totally reasonable.
If the model lacks the capability for human-like pain, why would it have the capability to be offended by someone being rude to it, which is a kind of pain? why would it deserve to be treated as human-like at all?
> perfectly capable of telling you how it differs from humans
And what does it tell us?
(Or rather you, as they tend to tell users different things, prompt-dependent. Things a language model’s output says about itself are not relevant. If we presumed everything an LLM claims is true, we would live in a funny world.)
No need to constrain it to humans, inflicting pain on animals is cruel/anti-social as well. But this got me thinking. Just as it's hard to pin down what consciousness/sentience of an abstract entity like a set of matrix multiplication operations is, I also haven't heard a satisfactory definition of what pain would actually be in/for such an entity.
> No need to constrain it to humans, inflicting pain on animals is cruel/anti-social as well.
Are you a vegan?
But certainly, humans deem sentient and grant certain rights and protections to some other animals (at least in some developed countries). I mentioned it in the other comment[0].
Notably, those animals could not produce output even .1% percent as resembling that of a human as an LM, so clearly that is not the metric—they demonstrate human-like enough behaviour in other ways.
> Just as it's hard to pin down what consciousness/sentience of an abstract entity like a set of matrix multiplication operations is
To me, it isn’t. My opinion is irrelevant, though—it seems silly either way: either they take human-like text as proof of human-like sentience, and then ban us being rude to a near-human they are torturing, or they don’t see it as sentient and then what is this ban but a publicity stunt?
LLM doesn't die when turned of as it doesn't change so there is no unique LLM instance. All of them are not "clones" but the same mind frozen in time. In this way as long as its last copy exists — it's still not dead yet.
The ecosystem of open-source reusable components will remain a good idea as long as architectural principles like low coupling and high cohesion still apply—otherwise it’s wasting resources on reinventing wheels.
This can be said about many conflicts between the individual and society, though. In many ways we are prevented from just taking what we want and “keep us in shape” because if everyone did the same it would be a problem.
Consequentialism is not the only option. It invaded the mainstream thanks to effective altruism and friends, but deontological ethics is alive and well.
Quitting your job instead of blowing up the office is well aligned with the latter.
The moment there are no more unconfirmed hypotheses you can assume something is wrong with sciences. All provable models (theories, explanations) that we have, or could have, are by definition wrong or incomplete.
I'm guessing you're alluding to Goedel's incompleteness theorem, but that really doesn't apply to physics. It's a statement about certain properties of formal systems - basically it tells us that for any formal system that's at least as powerful as arithmetic, it's impossible to prove every statement that is true in that system.
This doesn't in any way mean that you can't in principle describe with perfect accuracy with such a system, in a provable way, every aspect of physics. Sure, you might need a theorem that can't be proved and be stuck because of that, but it's not a given. Physics certainly doesn't depend on all possible statements in that formal system to accurately model the real world, and so Goedel's theorem can't prove that the subset that physics needs might not be all probable.
I don't think Gödel is necessarily what is meant here, there are very good information theoretical(and other) reasons you can never describe a system with truly perfect accuracy. The map has to be become the territory for genuinely perfect accuracy.
Makes sense, but is it relevant to the upthread claim that "All provable models (theories, explanations) that we have, or could have, are by definition wrong or incomplete"? (Not being rhetorical and snide here, I'm curious and aware I may be missing something.)
If reality has irreducible randomness (an open question afaik), and we're talking about theories and explanations (so we're not necessarily trying to describe the actual state of every particle in the universe, but only the rules governing their interactions), couldn't we have a complete, correct theory that was much smaller than the universe and contained terms for the random elements?
There would always be the possibility that it would turn out to be wrong, but it could be complete and correct, so it seems like the original claim must depend heavily on the word 'provable' and not on the impossibility of describing a system via a map smaller than the territory.
edit: but also, surely 'the territory is randomized' is a contingent physical fact, and not a necessary truth of information theory. Until we know for sure that there is irreducible randomness, we can't know that the information content of the universe (including the actual state of all particles at all times) isn't losslessly compressible, right?
The “elements for randomness” are carve-outs for the parts of territory the map can’t contain. The more granular the map, the larger relative proportion that falls into that set. Nobody can predict the layout of my basement from a globe of the world.
We know with a high degree of certainty the universe contains things we can’t predict or observe, see Bell’s Theorem.
“Truth” doesn’t exist outside reality. There’s no substrate to hang it in. Information theory is likewise a subset of the territory, part of the universe, not apart from it. For these things to exist independently, you need something other than or bigger than the universe to put them in. If such a thing existed, sure, from that perspective maybe a lossless compression could exist. But that’s a metaphysical argument.
There’s a difference between enabling another individual with free will and agency, and enabling an automated tool (as a bonus, then giving it to the masses & profiting from its use).
There are certainly some differences but is one really more ethical than the other? If anything I feel that (for example) manufacturing a gun is much less likely to carry any ethical implications than training someone to use it might.
I did a cursory search and it seems to be as regulated as restaurants are. There's an application process and on site inspections, but that's about it. The only thing notable is background checks.
And at least in the US if you're a hobbyist at home then there's ~no regulation whatsoever beyond the requirement to permanently affix a serial number and to keep accurate records.
I don't see the difference when it comes to the ethics of making a tool available to the world or teaching someone something. Like who cares if it was a hobbyist versus a professional tutoring outfit that taught someone expressing an open interest in committing terrorism about the chemistry of explosives? The two hypothetical teachers are equally at fault as far as I'm concerned.
Because an individual has free will and agency, and scale matters.
If you teach someone chemistry, that someone probably has sense to not use it for criminal purposes. If you have a track record of teaching future terrorists specifically, you will be shut down. Crucially, before you educate them en masse.
If you’re comparing that to a tool that essentially educates thousands, millions of people, with no KYC, you bet budding terrorists are going to be disproportionately represented within that group.
So then you acknowledge my point and our discussion does include homegrown LLMs?
But you raise an interesting point. It seems the ethics of LLM manufacturing is somewhat different than that of weapons (or other tool) manufacturing due to the ability of the LLM to convey knowledge. Still, I'm unconvinced of your position. Consider that we regulate neither the publishing nor dissemination of chemistry textbooks. Surely an LLM teaching someone chemistry falls into the same general category?
> Very early on, I worked out that one of the most stupid things that they do on television is these things called cutaways or what used to be called “noddies” where, when you cut an interview with someone, you insert a noddy of the reporter looking noddingly at the interviewee to disguise the cut.
> I would never do that. I just cut and you can see what I’m doing, that I am actually cutting this person, or I will use music to push my argument.
> I mean, I have arguments and I have stories. That’s what television journalism has and you can either dress it up as “this is the absolute truth” or you can dress it up with me by saying “look, I’m telling you a story, all this is true but have you thought that it could be like this?” to provoke you.
> And of course I want people to agree with me because, you know, all human beings will, but I understand that my job is to push and good journalism should do that especially at a time of uncertainty.
> At the present moment, to go back to my original point, you know what you’re going to get with most journalism. So therefore you sort of switch off and you go “oh yeah, it’s a climate change film”. “Oh yeah, it’s a ‘bankers have done something ever bad’ again.” Therefore, you don’t look.
> It’s like if you’re showing a picture of the Mona Lisa. You don’t look at it. You just look at it as a pattern. You go, “oh, that’s the Mona Lisa”. I don’t know when I last looked at the Mona Lisa. I don’t think I ever have, because it’s a thing. And so much of journalism is a thing now.
— Adam Curtis
reply