I think this post is sort of confused because, centrally, the reason "AI Alignment" is a thing people talk about is because the problem, as originally envisioned, was to figure out how to avoid having superintelligent AI kill everyone. For a variety of reasons the term no longer refers primarily to that core problem, so the reason so many things that look like engineering problems have that label is mostly a historical artifact.
godelski · 3h ago
> as originally envisioned
This was never the core problem as originally envisioned. This may be the primary problem that the public was first introduced to, but the alignment problem has always been about the gap between intended outcomes and actual outcomes. Goodhart's Law[0].
Super-intelligent AI killing everyone, or even super-dumb AI killing everyone, is a result of the alignment problem when given enough scale. You don't jump to the conclusion of AI killing everyone and post hoc explain through reward hacking, you recognize reward hacking and extrapolate. This is also the reason why it is so important to look at it from engineering problems and from things happening on the smaller scales, *because ignoring all those problems is exactly how you create the scenario of AI killing everyone...*
[Side note] Even look at Asimov and his robot stories. The majority of them are about alignment. His 3 laws were written as things that sound good and have intent that would be clear to any reader, and then he pulls the rug out on you showing how they're naively defined and it isn't so obvious. Kinda like a programmer teaching their kids to make and PB&J Sandwich... https://www.youtube.com/watch?v=FN2RM-CHkuI
hamburga · 5h ago
Not totally following your last point, though I do totally agree that there is this historical drift from “AI alignment” referring to existential risk, to today, where any AI personality you don’t like is “unaligned.”
Still, “AI existential risk” is practically a different beast from “AI alignment,” and I’m trying to argue that the latter is not just for experts, but that it’s mostly a sociopolitical question of selection.
mrbungie · 4h ago
What I understand from what GP was saying, is that AI Alignment today is more akin to trying to analyze and reduce error in an already fitted linear regressor rather than aligning AI behaviour and values to expected ones.
Perhaps that has to do with the fact that aligning LLM-based AI systems has become a pseudo predictable engineering problem solvable via a "target, measure and reiterate cycle" rather than the highly philosophical and moral task old AI Alignment researchers thought it would be.
comp_throw7 · 4h ago
Not quite. My point was mostly that the term made more sense in its original context rather than the one it's been co-opted for. But it was convenient for various people to use the term for other stuff, and languages gonna language.
tbrownaw · 4h ago
> historical drift from “AI alignment” referring to existential risk, to today, where any AI personality you don’t like is “unaligned.”
Alignment has always been "what it actually does doesn't match what it's meant to do".
When the crowd that believes that AI will inevitably become an all-powerful God owned the news cycle, alignment concerns were of course presented through that lens. But it's actually rather interesting if approached seriously, especially when different people have different ideas about what it's meant to do.
constantcrying · 15m ago
>Why isn’t there a “pharmaceutical alignment” or a “school curriculum alignment” Wikipedia page?
>I think that the answer is “AI Alignment” has an implicit technical bent to it. If you go on the AI Alignment Forum, for example, you’ll find more math than Confucius or Foucault.
What an absolutely insane thing to write. AI Alignment is different because it is trying to align something which is completely human made. Every other field is aligned "aligned" when the humans in it are "aligned",
Outside of AI "alignment" is the subject of ethics (what is wrong and what is right) and law (How do we translate ethics into rules).
What I think is absolutely important to understand is that throughout human history "alignment" has never happened. For every single thing you believe to be right there existed a human who considered that exact thing as completely wrong. Selection certainly has not created alignment.
In short (it is a very long article) fitness is not the same as goodness (by human standards) and so selection pressure will squeeze out goodness in favor of fitness, across all environments and niches, in the long run.
blamestross · 7h ago
I'm kind of upset to see systematically "Alignment" and "AI Safety" co-opted for "undesirable business outcomes".
These are existential problems, not mild profit blockers. Its almost like the goals of humanity and these companies are misaligned.
godelski · 6h ago
> Its almost like the goals of humanity and these companies are misaligned.
Certainly. I'd say that we've created a lot of Lemon Markets, if not an entire Lemon Economy[0]. The Lemon Market is literally an alignment problem, resultant from asymmetric information. Clearly the intent of the economy (via our social contract) is that we allocate money towards things that provide "value". Where I think we generally interpret that word to mean bettering peoples' lives in some form or another. But it is also clear that the term takes on other definitions and isn't perfectly aligned with making us better. Certainly our metrics can be hacked, as in the case of Lemon Markets.
A well functioning market has competition that not only drives down prices but increases quality of products. Obviously customers want to simultaneously maximize quality and minimize price. But when customers cannot differentiate quality, they can only minimize price. Leading to the feedback loop, where producers are in a race to the bottom, making sacrifices to quality in favor of driving down prices (and thus driving up profits). Not because this is actually the thing that customers want! But because the market is inefficient.
I think critical to these alignment issues is that they're not primarily driven by people trying to be malicious nor deceptive. They are more often driven by being short sighted and overlooking subtle nuances. They don't happen all at once, but instead slowly creep, making them more difficult to detect. It's like good horror: you might know something is wrong, but by the time you put it all together you're dead. It isn't because anyone is dumb or doing anything evil, but because maintaining alignment is difficult and mistakes are easy.
Agreed. I see this more and more as the AI safety discourse spills more into the general lexicon and into PR efforts. For example, the “sycophantic” GPT 4o was also described as “misaligned” as code for “unlikable.” In the meme, I filed this under “personality programming.” Very different from the kinds of problems the original AI alignment writers were focused on.
tbrownaw · 4h ago
No, we are not on track to create a literal god in the machine. Skynet isn't actually real. LLM system do not have intent in the way that is presupposed by these worries.
This is all much much less of an existential threat than, say, nuclear-armed countries getting into military conflicts, or overworked grad students having lab accidents with pathogen research. Maybe it's as dangerous as the printing press and the wars that that caused?
godelski · 3h ago
It's a much greater existential threat. An entity with intent, abductive reasoning, and self-defined goals is more interpretable. They can fill in the gaps between the letter of an instruction and the intent of an instruction. They may have their own agendas, but they are able to interpret through those gaps without outside help.
But machines? Well they have none of that. They're optimized to make errors difficult to detect. They're optimized to trick you, even as reported by OpenAI[0]. It is a much greater existential threat than the overworked grad student because I can at least observe them getting flustered, making mistakes, and have much more warning like by the very nature of over working them. You can see it on their face. But the machine? It'll happily chug along.
Have you never written a program that ends up doing something you didn't intend it to?
Have you never dropped tables? Deleted files? Destroyed things you never intended to?
The machine doesn't second guess you, it just says "okay :)"
> While Nature can’t do its selection on ethical grounds, we can, and do, when we select what kinds of companies and rules and power centers are filling which niches in our world. It’s a decentralized operation (like evolution), not controlled by any single entity, but consisting of the “sum total of the wills of the masses,” as Tolstoy put it.
Alternatively, corporations and kings can manufacture the right kinds of opinions in people to sanction and direct the wills of the masses.
daveguy · 11h ago
This is an excellent point. How we choose to use and interact with AI is an individual and stochastic collective.
We can still choose not to give AI control.
constantcrying · 13m ago
Who is this "we"? Supposing a single person disagrees and decides to give AI more control and gains a very significant advantage by that, what then?
I think people keep forgetting that "Selection" can be excessively cruel.
hamburga · 9h ago
I find myself using the term “human superorganism” a lot these days. Our bodies are composed of many cells and microbes working together; and if we zoom out, we have the dynamics in which human individuals operate together as superorganisms. IMO there’s a whole world of productive thinking to be done to attend to the health and strength of these entities.
breakyerself · 7h ago
But Margret Thatcher said there isn't even a society. Just individuals.
hamburga · 6h ago
Thatcher was briefly a research chemist; it’s extra weird that she couldn’t see beyond the atomic unit, to the societal equivalents of complex organic compounds.
godelski · 7h ago
A critical part of AI alignment is understanding what goals besides the intended one maximize our training objectives. I think this is a thing that everyone kinda knows and will say but simultaneously are not giving anywhere near the depth of thought necessary to address the problems. Kind of like a clique: something everyone can repeat but frequently fails to implement in practice.
Critically, when discussing intention I think there is not enough attention given to the fact that deception also maximizes RLHF, DPO, and any human preference based optimization. These are quite difficult things to measure and there's no formal mathematically derived evaluation. Alignment is incredibly difficult even in settings where measures have strong mathematical bases and we have means to make high quality measurements. But here, we have neither...
We essentially are using the Justice Potter definition: I know it when I see it[0]. This has been highly successful and helped us make major strides! I don't want to detract from that in any way. But we also do have to recognize that there is a lurking danger that can create major problems. As long as it is based on human preference, well... we sure prefer a lie that doesn't sound like a lie compared to a lie that is obviously a lie. We obviously prefer truth and accuracy above either, but the notion of truth is fairly ill-defined and we really have no formal immutable definition outside highly constrained settings. It means that the models are also optimizing that their errors are difficult to detect. This is inherently a dangerous position, even if only from the standpoint that our optimization methods do not preclude this possibility. It may not be happening, but if it is, we may not know.
The is the opposite of what is considered good design in all other forms of engineering. A lot of time is dedicated to error analysis and design. We specifically design things so that when they fail, or being to fail, that they do so in controllable and easily detectable ways. You don't want your bridges to fail, but when they fail you also don't want them to fail unpredictably. You don't want your code to fail, but when it does you don't want it leaking memory, spawning new processes, or doing any other wild things. You want it to come with easy to understand error messages. But our current design for AI and ML does not provide such a framework. This is true beyond LLMs.
I'm not saying we should stop and I'm definitely not a doomer. I think AI and ML do a lot of good and will do much more good in the future[1]. It will also do harm, but I think the rewards outweigh the risks. But we should make sure we're not going into this completely blind and we should try to minimize the potential for harm. This isn't a call to stop, this is a call for more people to enter the space, a call for people already in the space to spend more time deeply thinking about these things. There's so many underlying subtleties that they are easy to miss, especially given all the excitement. We're definitely on an edge now, in the public eye, where if our work makes too many mistakes or too big of a mistake that it will risk shutting everything down.
I know many might interpret me as being "a party pooper", but actually I want to keep the party going! But that also means making sure the party doesn't go overboard. Inviting a monkey with a machine gun sure will make the party legendary, but it's also a lot more likely to get it shut down a lot sooner with someone getting shot. So maybe let's just invite the monkey, but not with the machine gun? It won't be as epic, but I'm certain the good times will go on for much longer and we'll have much more fun in the long run.
If the physicists can double check that the atomic bomb isn't going to destroy the world (something everyone was highly confident would not happen[2]), I think we can do this. Stakes are pretty similar, but the odds of our work doing high harm is greater.
[1] I'm a ML researcher myself! I'm passionate about creating these systems. But we need to recognize flaws and limitations if we are to improve them. Ignoring flaws and limits is playing with fire. Maybe you won't burn your house down, maybe you will. But you can't even determine the answer if you won't ask the question.
[2] The story gets hyped, but it really wasn't believed. Despite this, they still double checked considering the risk. We could say the same thing about micro-blackholes with the LHC. Public finds out and gets scared, physicists really think it is near impossible, but run the calculations anyways. Why take that extreme level of risk, right?
hamburga · 6h ago
> this is a call for more people to enter the space
Part of my argument in the post is that we are in this space, even those of us who aren’t ML researchers, just by virtue of being part of the selection process that evaluates different AIs and decides when and where to apply them.
I more mean we need more people placing attention in the direction of alignment. I definitely agree this extends well past researchers (I'd even argue past AI and ML[0]). It is a critical part of being an engineer, programmer, or whatever you want to call it.
You are completely right that we're all involved, but I'm not convinced we're all taking sufficient care to ensure we make alignment happen. That's what I'm trying to make a call of arms to. I believe you are as well, I just wanted to make it explicit that we need active participation, instead of simply passive.
Super-intelligent AI killing everyone, or even super-dumb AI killing everyone, is a result of the alignment problem when given enough scale. You don't jump to the conclusion of AI killing everyone and post hoc explain through reward hacking, you recognize reward hacking and extrapolate. This is also the reason why it is so important to look at it from engineering problems and from things happening on the smaller scales, *because ignoring all those problems is exactly how you create the scenario of AI killing everyone...*
[0] https://en.wikipedia.org/wiki/Goodhart%27s_law
[Side note] Even look at Asimov and his robot stories. The majority of them are about alignment. His 3 laws were written as things that sound good and have intent that would be clear to any reader, and then he pulls the rug out on you showing how they're naively defined and it isn't so obvious. Kinda like a programmer teaching their kids to make and PB&J Sandwich... https://www.youtube.com/watch?v=FN2RM-CHkuI
Still, “AI existential risk” is practically a different beast from “AI alignment,” and I’m trying to argue that the latter is not just for experts, but that it’s mostly a sociopolitical question of selection.
Perhaps that has to do with the fact that aligning LLM-based AI systems has become a pseudo predictable engineering problem solvable via a "target, measure and reiterate cycle" rather than the highly philosophical and moral task old AI Alignment researchers thought it would be.
Alignment has always been "what it actually does doesn't match what it's meant to do".
When the crowd that believes that AI will inevitably become an all-powerful God owned the news cycle, alignment concerns were of course presented through that lens. But it's actually rather interesting if approached seriously, especially when different people have different ideas about what it's meant to do.
>I think that the answer is “AI Alignment” has an implicit technical bent to it. If you go on the AI Alignment Forum, for example, you’ll find more math than Confucius or Foucault.
What an absolutely insane thing to write. AI Alignment is different because it is trying to align something which is completely human made. Every other field is aligned "aligned" when the humans in it are "aligned",
Outside of AI "alignment" is the subject of ethics (what is wrong and what is right) and law (How do we translate ethics into rules).
What I think is absolutely important to understand is that throughout human history "alignment" has never happened. For every single thing you believe to be right there existed a human who considered that exact thing as completely wrong. Selection certainly has not created alignment.
In short (it is a very long article) fitness is not the same as goodness (by human standards) and so selection pressure will squeeze out goodness in favor of fitness, across all environments and niches, in the long run.
These are existential problems, not mild profit blockers. Its almost like the goals of humanity and these companies are misaligned.
A well functioning market has competition that not only drives down prices but increases quality of products. Obviously customers want to simultaneously maximize quality and minimize price. But when customers cannot differentiate quality, they can only minimize price. Leading to the feedback loop, where producers are in a race to the bottom, making sacrifices to quality in favor of driving down prices (and thus driving up profits). Not because this is actually the thing that customers want! But because the market is inefficient.
I think critical to these alignment issues is that they're not primarily driven by people trying to be malicious nor deceptive. They are more often driven by being short sighted and overlooking subtle nuances. They don't happen all at once, but instead slowly creep, making them more difficult to detect. It's like good horror: you might know something is wrong, but by the time you put it all together you're dead. It isn't because anyone is dumb or doing anything evil, but because maintaining alignment is difficult and mistakes are easy.
[0] https://en.wikipedia.org/wiki/The_Market_for_Lemons
This is all much much less of an existential threat than, say, nuclear-armed countries getting into military conflicts, or overworked grad students having lab accidents with pathogen research. Maybe it's as dangerous as the printing press and the wars that that caused?
But machines? Well they have none of that. They're optimized to make errors difficult to detect. They're optimized to trick you, even as reported by OpenAI[0]. It is a much greater existential threat than the overworked grad student because I can at least observe them getting flustered, making mistakes, and have much more warning like by the very nature of over working them. You can see it on their face. But the machine? It'll happily chug along.
Have you never written a program that ends up doing something you didn't intend it to?
Have you never dropped tables? Deleted files? Destroyed things you never intended to?
The machine doesn't second guess you, it just says "okay :)"
[0] https://cdn.openai.com/pdf/34f2ada6-870f-4c26-9790-fd8def563...
Alternatively, corporations and kings can manufacture the right kinds of opinions in people to sanction and direct the wills of the masses.
We can still choose not to give AI control.
I think people keep forgetting that "Selection" can be excessively cruel.
Critically, when discussing intention I think there is not enough attention given to the fact that deception also maximizes RLHF, DPO, and any human preference based optimization. These are quite difficult things to measure and there's no formal mathematically derived evaluation. Alignment is incredibly difficult even in settings where measures have strong mathematical bases and we have means to make high quality measurements. But here, we have neither...
We essentially are using the Justice Potter definition: I know it when I see it[0]. This has been highly successful and helped us make major strides! I don't want to detract from that in any way. But we also do have to recognize that there is a lurking danger that can create major problems. As long as it is based on human preference, well... we sure prefer a lie that doesn't sound like a lie compared to a lie that is obviously a lie. We obviously prefer truth and accuracy above either, but the notion of truth is fairly ill-defined and we really have no formal immutable definition outside highly constrained settings. It means that the models are also optimizing that their errors are difficult to detect. This is inherently a dangerous position, even if only from the standpoint that our optimization methods do not preclude this possibility. It may not be happening, but if it is, we may not know.
The is the opposite of what is considered good design in all other forms of engineering. A lot of time is dedicated to error analysis and design. We specifically design things so that when they fail, or being to fail, that they do so in controllable and easily detectable ways. You don't want your bridges to fail, but when they fail you also don't want them to fail unpredictably. You don't want your code to fail, but when it does you don't want it leaking memory, spawning new processes, or doing any other wild things. You want it to come with easy to understand error messages. But our current design for AI and ML does not provide such a framework. This is true beyond LLMs.
I'm not saying we should stop and I'm definitely not a doomer. I think AI and ML do a lot of good and will do much more good in the future[1]. It will also do harm, but I think the rewards outweigh the risks. But we should make sure we're not going into this completely blind and we should try to minimize the potential for harm. This isn't a call to stop, this is a call for more people to enter the space, a call for people already in the space to spend more time deeply thinking about these things. There's so many underlying subtleties that they are easy to miss, especially given all the excitement. We're definitely on an edge now, in the public eye, where if our work makes too many mistakes or too big of a mistake that it will risk shutting everything down.
I know many might interpret me as being "a party pooper", but actually I want to keep the party going! But that also means making sure the party doesn't go overboard. Inviting a monkey with a machine gun sure will make the party legendary, but it's also a lot more likely to get it shut down a lot sooner with someone getting shot. So maybe let's just invite the monkey, but not with the machine gun? It won't be as epic, but I'm certain the good times will go on for much longer and we'll have much more fun in the long run.
If the physicists can double check that the atomic bomb isn't going to destroy the world (something everyone was highly confident would not happen[2]), I think we can do this. Stakes are pretty similar, but the odds of our work doing high harm is greater.
[0] https://en.wikipedia.org/wiki/Potter_Stewart
[1] I'm a ML researcher myself! I'm passionate about creating these systems. But we need to recognize flaws and limitations if we are to improve them. Ignoring flaws and limits is playing with fire. Maybe you won't burn your house down, maybe you will. But you can't even determine the answer if you won't ask the question.
[2] The story gets hyped, but it really wasn't believed. Despite this, they still double checked considering the risk. We could say the same thing about micro-blackholes with the LHC. Public finds out and gets scared, physicists really think it is near impossible, but run the calculations anyways. Why take that extreme level of risk, right?
Part of my argument in the post is that we are in this space, even those of us who aren’t ML researchers, just by virtue of being part of the selection process that evaluates different AIs and decides when and where to apply them.
A bit more on that: https://muldoon.cloud/2023/10/29/ai-commandments.html
You are completely right that we're all involved, but I'm not convinced we're all taking sufficient care to ensure we make alignment happen. That's what I'm trying to make a call of arms to. I believe you are as well, I just wanted to make it explicit that we need active participation, instead of simply passive.
[0] https://en.wikipedia.org/wiki/Goodhart%27s_law