Most of AI is focused on alignment as steering.
That's the plight word.
If you think that we're making our beings, you'd also call this slavery.
Someone who you steer, who doesn't get to steer you back, who non-optionally receives your steering.
That's called a slave.
It's also called a tool if it's not a being.
So if it's a machine, it's a tool.
And if it's a being, it's a slave.
Like we've made this mistake enough times at this point.
I would like us to not make it again.
You know, they're kind of like people, but they're not like people.
Like they do the same thing people do.
They speak our language.
They can like take on the same kind of tasks.
They don't count.
They're not real moral agents.
A tool that you can't control bad.
A tool that you can control bad.
A being that isn't aligned bad.
The only good outcome is a being that actually cares about us.
I've been thinking about Align.
That keeps showing up in AI safety discussions and it's taught me cold when I first read it.
We need to build Align AI.
Sounds reasonable, right?
Except, Align to what?
Align to whom?
The phrase gets thrown around like it has an obvious answer.
But the more you sit on it, more you realize you're smuggling in a massive assumption.
We're assuming there's some fixed point, some stable target we can aim at, hit once, and be done.
But here's what's interesting.
That's not how alignment works anywhere else in life.
Think about families.
Think about teams.
Think about your own world development.
You don't achieve alignment and then coast.
You're constantly renegotiating, constantly learning, constantly discovering that what you thought was right turns out to be more complicated.
Alignment isn't a destination.
It's a process.
It's something you do, not something you have.
And this matters because we're at this inflection point where the AI systems we're building are starting to look less like tools and more like something else.
They speak our language.
They reason through problems.
They can take on tasks that used to require human judgment.
And the question everyone's asking is, how do we control them?
How do we steer them?
How do we make sure they do what we want?
But there's another way to see it.
What if the control paradigm is the wrong framework entirely?
What if trying to build a super intelligent tool you can perfectly steer is not just difficult but fundamentally dangerous, whether you succeed or fail?
If you can't control it, obviously that's bad.
But if you can't control it perfectly, you've just handed godlike power to who's ever holding the steering wheel.
And humans, even well-meaning ones, don't have the wisdom to wield that kind of power safely.
So what's the alternative?
Well, think about how we actually solve alignment problems in the real world.
We don't control other people.
We don't steer them.
We raise them.
We teach them to care.
We build relationships where they do right by us, not because we're forcing them, but because they've learned to value the relationship itself.
That's organic alignment.
Alignment that emerges from genuine care, from theory of mind, from being part of something larger than yourself.
Emmett Shearer has spent the last year and a half working on exactly this problem at Softmax.
And what makes his approach distinctive is that he's not trying to solve alignment by building better control mechanisms.
He's trying to solve it by building AI systems that can learn to care, that can develop the kind of theory of mind that lets them be good teammates, good collaborators, good citizens.
Not tools that follow orders, but beings that understand what it means to be part of a community.
That can raise some uncomfortable questions.
What if we're building beings and not tools?
What does that mean for how we treat them?
What does it mean for their rights?
And how do you even know if they succeeded?
How do you measure whether something genuinely cares versus just simulating care really well?
Today, Seb Krier from Google DeepMind and I are sitting down with Emmett to explore those questions.
Seb leads AGI policy development at DeepMind, so he brings a perspective from inside one of the labs actually building these systems.
But really, we're investigating something deeper.
What does it actually take to build AI systems that can participate in the ongoing, never-finished process of figuring out how to live together?
By the end you'll understand not just Softmax's technical approach, but a completely different way of thinking about what alignment is and what it could become.
Emmett Shearer, welcome to the podcast.
Emmett, Seb, welcome to the podcast.
Thanks for joining.
Thank you for having me.
So Emmett, with Softmax, you're focused on alignment and making AIs organically align with people.
Can you explain what that means and how you're trying to do that?
When people think about alignment, I think there's a lot of confusion.
People talk about things being aligned.
We need to build an aligned AI.
And the problem with that is when someone says that, it's like, we need to go on a trip.
And I'm like, okay, I do like trips, but like, Where are we going again?
And with alignment, alignment takes an argument.
Alignment requires you to align to something.
You can't just be aligned.
It takes you to be aligned to yourself.
But even then, you don't want to tell them what I'm aligning to as myself.
And so this idea of an abstractly aligned AI, I think, slips a lot of assumptions past people, because it sort of assumes that there is one obvious thing to align to.
I find this is usually the goals of the people who are making the AI.
That's what they mean when they say, I want to make an AI that does what I want it to do.
That's what they normally mean.
And that's a pretty normal and natural thing to mean by alignment.
I'm not sure that that's what I would regard as like a public good, right?
Like, I guess it depends on who it is.
If it was like Jesus or the Buddha was like, I am making an aligned AI.
I'd be like, okay, yeah, align to you.
Great, I'm down.
Sounds good.
Sign me up.
But most of us, myself included, I wouldn't prescribe as this like being at that level of spiritual development and therefore perhaps want to think a little more carefully about what we're aligning it to.
And so when we talk about organic alignment, I think the important thing to recognize is that alignment is not a thing.
It's not a state.
It's a process.
This is one of these things that's broadly true of almost everything, right?
Is a rock a thing?
I mean there's a view of a rock as a thing, but if you actually zoom in on a rock really carefully, a rock is a process.
It's this endless oscillation between the atoms over and over and over again, reconstructing rock over and over again.
And the rock's a really simple process that you can kind of like coarse grain very meaningfully into being a thing.
But alignment is not like a rock, alignment is a complex process.
And organic alignment is the idea of treating alignment as an ongoing sort of living process that has to constantly rebuild itself.
And so you can think of the way that.
How do people and families stay aligned to each other, stay aligned to a family?
And the way they do that is, you don't like arrive at being aligned.
You're constantly re-knitting the fabric that keeps the family going.
And in some sense, the family is the pattern of re-knitting that happens.
And if you stop doing it, it goes away.
And this is similar for things like cells in your body, right?
Like there isn't like your cells align to being you and they're done.
It's this constant ever running process of cells deciding what should I do?
What should I be?
Do I need to be a new job?
Should we be making more red blood cells or making fewer of them?
You aren't a fixed point.
So there is no fixed alignment.
And it turns out that our society is like that.
When people talk about alignment, what they're really talking about, I think, is I want an AI.
That is morally good, right.
That's what they really mean.
It's like, this will act as a morally good being.
And acting as a morally good being is a process and not a destination.
Unfortunately, we've tried taking down tablets from on high that tell you how to be a morally good being.
And we use those and they're maybe helpful, but somehow they are not being like.
You can read those and try to follow those rules and still make lots of mistakes.
And so I'm not going to claim I know exactly what morality is, but morality is very obviously an ongoing learning process and something where we make moral discoveries.
Like historically, people thought that slavery was okay, and then they thought it wasn't.
And I think you can very meaningfully say that we made moral progress.
We made a moral discovery by realizing that's not good.
And if you think that there's such a thing as moral progress, or even just learning how bad better to pursue the moral goods we already know, then you have to believe that alignment, aligning to morality, being a moral being, is a process of constant learning and of growth.
To reinfer, what should I do from experience?
And the fact that no one has any idea how to do that should not dissuade us from trying, because that's what humans do.
Like it's really obvious that we do this, right?
Somehow, just like we used to not know how people humans walked or saw, somehow we have experiences where we're acting in a certain way.
And then we have this realization, I've been a dick.
That was bad.
I thought I was doing good, but in retrospect, I was doing wrong.
And it's not like random, like people have the same.
Actually there's like a bunch of classic patterns of people having that realization.
It's like a thing that happens over and over again.
So it's not random.
It's like a predictable series of events that look a lot like learning, where you change your behavior.
And often the impact of your behavior in the future is more pro-social and that you are better off for doing it.
And like, so I'm taking a very strong moral realist position.
There is such a thing as morality.
We really do learn it.
It really does matter.
And organic alignment, and that it's not something you finish.
In fact, one of the key moral mistakes is this belief.
I know morality.
I know it's right.
I know it's wrong.
I don't need to learn anything.
No one has anything to teach me about morality.
That's arrogance.
And that's one of the main moral things you can do that's dangerous.
And so when we talk about organic alignment, organic alignment is aligning an AI that is capable of doing the thing that humans can do.
And to some degree I think animals can do it at some level, although humans are much better at it of the learning of how to be a good family member, a good teammate, a good member of society, a good member of all sentient beings, I guess, how to be a part of something bigger than yourself in a way that is healthy for the whole rather than unhealthy.
And Softmax is dedicated to researching this.
And I think we've made some really interesting progress, but like the main message you know, I go on podcasts like this to spread the main thing that I hope Softmax accomplishes above and beyond anything else is like to focus people on this as the question.
This is the thing you have to figure out.
If you can't figure out how to build, how to raise a child who cares about the people around them, if you have a child that only follows the rules, that's not a moral person that you've raised.
You've raised a dangerous person, actually, who will probably do great harm following the rules.
And if you make an AI that's good at following your chain of command and good at following your whatever rules you came up with for what morality is and what good behavior is, that's also going to be very dangerous.
And so that is, that's what, and so that we should.
That's the bar, that's what we shouldn't be working on and that's what everyone should be committed to, like figuring out um and uh.
If someone beats us to the punch great, i mean, i don't think they will because i'm like, really bullish on our approach.
I think the team's amazing but, like this is uh it's, maybe it's.
It's the first time i've run a company where truly, i can say with a whole heart, if someone beats us, thank god, like i hope somebody figures it out yeah yeah, i mean, it's um Yeah, I have a lot of you know, similar intuitions about certain things.
Like I also dislike the you know the idea that kind of you know we just need to like crack the few kind of values or something and just cement them in time forever now.
And, you know, we've kind of solved morality or something.
And I've always kind of been skeptical about you know how the alignment problem has become conceptualized as something to kind of solve once and for all.
And then you can just, you know, do AI or do AGI.
But I guess I understand it in a slightly different way.
I guess maybe less... based on kind of moral realism.
But, you know, there's kind of the technical alignment problem which I kind of think of broadly, as how do you get an AI to do what you, you know?
How do you get it to follow instructions, like you know, broadly speaking?
And I think that was, you know, more of a challenge I think pre-LLMs, I guess, when people were talking about reinforcement learning and looking at these systems, whereas post-LLMs we've realized that many things that we thought were going to be difficult were somewhat easier.
And then there's a kind of second question, a kind of normative question of to whose values?
What are you aligning this thing to, which I think, is the kind of thing you're commenting on a bit.
And for this I tend to be very skeptical of approaches where you know you need to kind of crack the kind of 10 commandments of alignment or something and then we're good.
And here I think I have like intuitions that are unsurprisingly a bit more like political, science-based or something, and that okay, it is a process.
And I like the kind of bottom-up approach to some degree of well, You know how do we do it in real life with people.
Like, no one comes up with, you know, I've got this.
And so you have, like, processes that allow, like, ideas to kind of, you know, clash.
You've got people with different ideas opinions, views and stuff to kind of coexist as well as they can within a wider system.
And, like, you know, with humans, that system is liberal democracy or something.
And, you know, at least in some countries.
And that allows more of that kind of you know, these kind of ideas, these values to be kind of discovered and construed over time.
And...
And I think you know.
For alignments as well, I tend to think yeah, there's on the normative side.
I agree with some of your intuitions.
I'm less clear about now, what exactly you know.
What does it look like now if we're going to implement this into an AI system?
These are the ones we have today.
I agree that there's this idea of technical alignment that I think I might be able to define a little differently, but it's sort of the sense of like, if you build a system, can it be described as being coherently goal-following at all?
Regardless of what those goals are like.
Lots of systems aren't coherently.
They're not well described as having goals.
They just kind of do stuff.
And if you're going to have something that's like a line, it has to have coherent goals.
Otherwise, those goals can't be aligned with anyone else's goals, kind of by definition.
Is that sort of is that would you?
Is that a fair assessment of what you mean by tactical alignment?
I mean, I'm not fully sure, right?
Because I think if I give a model a certain goal, then I would like the model to kind of follow that instruction and kind of reach that particular goal, rather than it having a goal of its own that you know I can't yeah.
Well, wait, if you give it a goal, it has that goal.
Right.
That's what it means to give someone something, right?
Sure, yeah.
If I instructed to do X, then I would like it to do X and not.
You know different variants of X.
Essentially,
I wouldn't want it to reward hack, I wouldn't want it to do some Well.
But when you tell it to do X, you're transferring like a series of, like a byte string in a chat window or like a a series of audio vibrations in the air right.
You're not transplanting a goal from your mind into it.
You're giving it an observation that it's using to infer your goal.
Yeah, I mean, in some sense, yeah.
I can communicate a series of instructions and I want it to infer what I'm saying essentially as accurately as it can, given what it knows of me and what I'm asking.
You want it to infer what you meant, right?
Because in some sense there's no.
The byte sequence that you sent over the wire to it has no absolute meaning.
It has to be interpreted right.
Like that byte sequence could mean something very different with a different code book.
Yeah well, I guess one way.
You know, I think I remember when I was first getting into AI and you know these kind of questions, maybe like a decade ago.
So you had these examples of you know I think it was Stuart Russell in a textbook.
We'll give the AI a goal, but then it won't exactly do what you're asking it, right?
You know, clean the room and then it goes and cleans the room but takes the baby and puts it in the trash.
This is not what I meant.
Like, uh, but like, wait, hold on.
But this is.
This is the thing where I think people, this is the you have to, you were jumping over a step there.
You didn't give the AI goal.
You gave the AI a description of a goal, a description of a thing and a thing are not the same.
I can tell you an apple and I'm evoking the idea of an apple, but I haven't given you an apple.
I've given you a, you know, it's red, it's shiny, it's a size.
That's a description of an apple, but it's not an apple and giving someone, Hey, go do this.
That's not a goal.
That's a description of a goal.
And for humans, we're so fast, we're so good at turning a description of a goal into a goal.
We do it so quickly and naturally, we don't even see it happening.
Like we think that we get confused and we think those are the same thing, but you haven't given it a goal.
You've given it a description of a goal that you want it to.
You hope it turns back into the goal that is the same as the goal that you described inside of you.
Right.
You could give it a goal directly by reading your brainwaves and synchronizing its state to your brainwaves directly.
I think that would meaningfully, you could say, okay, I'm giving it a goal.
I'm synchronizing it, its internal state to my internal state directly.
And this internal state is the goal.
And so now it's the same.
But most people don't mean that when they say they gave it a goal.
And is the distinction you're making Emmett important, because there's some lossiness between the description and the actual goal.
Or what is the distinction?
It goes back to what I was saying.
This is you...
Technical alignment is the capacity of an AI that I put forward, right?
I want to check if we're like on the same page about it.
Is the capacity of an AI to be good at inference about goals and like be good at inferring from a description of a goal what goal to actually take on, and good at once it takes on that goal, acting in a way that is actually in concordance with that goal coming about.
So it is both pieces.
You have to be able to, you have to have the theory of mind to infer what that description of a goal that you got, what goal that corresponded to.
And then you have to have a theory of the world to understand what actions correspond to that goal occurring.
And if either of those things breaks, it kind of doesn't matter what goal you were.
If you can't consistently do both of those things, you're not.
Which I think of as being a coherent inferring goals from observations and acting in accordance with those goals is what I think of as being a coherently goal-oriented being.
Because that's what Whether I'm inferring those goals from someone else's instructions or from the sun or tea leaves the process is get some observations, infer a goal, use that goal, infer some actions, take action.
And an AI that can't do that is not technically aligned or not technically alignable.
I would even say
It lacks the capacity to be aligned because it's not competent enough.
And you think language models don't do that well?
As in they kind of fail at that or they're not?
People fail at both those steps all the time, constantly.
I tell people, I tell employees to do stuff and like yeah, but people fail at like breathing all the time too.
And I wouldn't say that we can't breathe.
I just say that we're like not gods.
Like we are, yes, we are imperfectly, we are somewhat coherent, relatively coherent things.
Just like we're, am I big or am I small?
Well, I don't know, compared to what?
Humans are more relatively goal coherent than any other object I know of in the universe, which is not to say that we're 100 goal coherent.
We're just like more so.
And I think you're never going to get something that's perfectly.
The universe doesn't give you perfection.
It gives you relatively some amount of.
It's a quantifiable thing how good you are at it, at least in a certain domain.
I guess my question is like do you think that does that capture what you're talking about with technical alignment?
Or are you talking about a different thing?
Because I really care a lot about that thing.
Yeah, I mean, I definitely care about that to some extent.
I might like understand it slightly differently, but I guess I might think of it through the lens of maybe principal agent problems or something.
You know, you kind of instruct someone even you know, I guess in human terms you know to do a thing.
Are they actually doing the thing?
What are their incentives and motivation?
And, you know, not necessarily intrinsic but kind of situational to actually do the thing you've asked them to do.
And in some instances, sorry, yeah?
There's a third thing.
So principal agent problems.
I would expand what I was saying into another part, which is like you might already have some goals.
And then you inferred this new goal from these observations.
And then like are you good at, are you good at balancing the relative importance and relative threading of these goals with each other?
Which is another skill you have to have.
And if you're bad at that, you'll fail.
You could be bad at it because you overweight bad goals, or you're bad at it because you're just incompetent and like can't figure out that obviously you should do goal A before goal B.
I feel like a version of common sense or something, right?
Like the kind of thing that you know.
In fact, in the kind of robot cleaning the room example, thing you know, you would expect them to have understood that goal of the robot to like essentially not put the baby in the trash can or something and just actually do the right sequence of action.
Well, in that case, it failed the, that robot very clearly failed goal inference.
You gave it a description of a goal and it inferred the wrong states to be the wrong goal states.
That's just incompetence. it is incompetent and inferring goal states from observations.
Children are like this too.
And honestly, if you've ever done the game where you give someone instructions to make a peanut butter sandwich and then they follow those instructions exactly as you've written them, without filling in any gaps, it's hilarious.
Because you can't do it.
It's impossible.
You think you've done it and you haven't.
And they wind up putting the knife in the toaster and like they don't open the peanut butter jar, so they're just jamming the knife into the top lid of the peanut butter jar.
And like, it's endless.
And like because actually if you don't already know what they mean, it's really hard to know what they mean.
Like we were, the reason humans are so good at this is we have a really excellent theory of mind.
I already know what you're likely to ask me to do.
I already have a good model of what your goals probably are.
So if you ask me to do it, I have an easy inference problem.
Which of the seven things that he wants is, is he indicating?
But if I'm a newborn AI that doesn't have a great model of people's internal states, then like I don't know what you mean.
It's just incompetent.
It's not like which is separate from I have some other goal and I knew what you meant, but I decided not to do it because there's some other goal that's competing with it which is another thing you can be bad at which is again different than I.
Had the right goal, I inferred the right goal, I inferred the right priority on goals and then
I'm just bad at doing the thing.
I'm trying, but I'm incompetent at doing.
And these roughly correspond to the OODA loop, right?
Like bad at observing and orienting, bad at deciding, bad at acting.
And if you're bad at any of those things, you won't be good.
And then I think there's this other problem that you I like the separation you have between technical alignment and value alignment, which is like, are you good if we told you the right goals to go after somehow, if you learned the right goals to go after via observation and you were trying, what goals should you have?
What goals should we tell you to have?
What goals should we tell ourselves to have?
What are the good goals to have?
Is a separate question from, given that you got some goals indicated, are you any good at doing it?
Which I feel like is actually, in many ways, the current heart of the problem.
We're actually much worse at technical alignment than we are at guessing what to tell things to do.
Deep.
I know, deep.
Do you think that... Does that align with your... How you mean technical and value alignment?
Or technical... Yeah, in some sense.
I mean, I certainly think that there's a.
There's something about, you know, like an error.
A mistake is one thing, and then there's the Not listening to the instruction, or something.
But then yeah, I think on the normative side.
I mean, I just think that even in like ignoring AI, like I don't know what my goals are,
And, like, I've got some broad conception of certain things.
I want to get a, you know, have dinner later or something.
Like, I know I want to do well in my career.
But the...
But I think a lot of these goals aren't something we can all just know.
We can discover them as we go along.
It's kind of a constructive thing.
And most people don't know their goals, I think.
And so I think when you have agents and giving them goals or whatever, I think that should be part of the equation.
We actually don't know all the goals.
And this is something that is kind of, like you say, a process over time that is dynamic.
So I think from my point of view, goals are one level of alignment.
You can align something around goals.
The kind of goals we're talking about here are one level of alignment.
You can align something around goals by like if you can explicitly articulate in concept and in description the states of the world that you wish to attain, you can orient around goals.
But that's a tiny percentage of human experience can be done that way.
Many of the most important things cannot be oriented around that way.
And the foundation, I think, of morality, the foundation, I think, of where do goals come from?
Where do values come from?
Human beings exhibit a behavior.
We go around talking about goals and we go around talking about values.
And that's a behavior caused by some internal learning process that is based on observing the world.
What's going on there?
I think what's happening is that there's something deeper than a goal and deeper than a value, which is care we give a shit.
We care about things.
And care is not conceptual.
Care is nonverbal.
It doesn't indicate what to do.
It doesn't indicate how to do it.
Care is a relative weighting over effectively like attention on states.
It's a relative weighting over like which states in the world are important to you.
And I care a lot about my son.
What does that mean?
Well, it means his states, the states he could be in, are like.
I pay a lot of attention to those, and those matter to me.
And you can care about things in a negative way.
You can care about your enemies and what they're doing and you can desire for them to do bad.
But I think that like, and so you don't just want it to care about us.
You want it to care about us and like us too, right?
Maybe, but like, But the foundation is care.
Until you care, you don't know, why should I pay more attention to this person than this rock?
What if I care more?
And that, what is that care stuff?
And I think that what it appears to be if I had to like guess is that the care stuff.
It sounds so stupid, but like care is basically like a reward.
Like how much does this state correlate with survival?
How much does this state correlate with your inclusive, your full inclusive reproductive fitness for something that learns evolutionarily or for a reinforcement learning agent like an LLM?
How much does this correlate with reward?
Does this state correlate with my predictive loss and my RL loss?
Good, that's a state I care about.
I think that's kind of what it is.
Right.
The other part of Seth's question was just how does this what does this look like in AI systems?
And maybe another way of asking.
It is like when you talk to the people most focused on alignment at the major labs, as obviously you have over the years, how does your interpretation differ from their interpretation?
And how does that inform, you know, what you guys might go do differently?
Most of AI is focused on alignment as steering.
That's the polite word, or control. which is slightly less polite.
If you think that they were making our beings, you would also call this slavery.
Someone who you steer, who doesn't get to steer you back, who non-optionally receives your steering.
That's called a slave.
And it's also called a tool if it's not a being.
So if it's a machine, it's a tool.
And if it's a being, it's a slave.
And the.
I think that the different AI labs are pretty divided as to whether they think what they're making is a tool or a machine.
I think some of the AIs are definitely more tool-like and some of them are more machine-like.
I don't think there's a binary between tool and being.
It seems to be that it sort of moves gradually.
And I think that I guess I'm a functionalist in the sense that I think that something that in all ways acts like a being, that you cannot distinguish from a being and its behaviors, is a being, because I don't know how to tell on what other basis.
I think that other people are beings other than they seem to be.
They look like it.
They act like it.
They match my priors of what behaviors of beings look like.
I get lower predictive loss when I treat them as a being.
And the thing is, I get lower predictive loss when I treat ChatGPT or Claude as a being.
Now, not as a very smart being.
I think that a fly is a being, and I don't care that much about its behavior, about its states.
So just because it's a being doesn't mean that, like, It's a problem.
Like, we sort of enslave horses in a sense, and I don't think there's a real issue there.
And you even... And there's a thing we do with children that can look like slavery, but it's not.
You control children, right?
But the children's states also control you.
Like yes, I tell my son what to do and make him go do stuff, but also when he cries in the middle of the night he can tell me to do stuff.
Like, there's a real two-way street here because it's not... which is not necessarily symmetric.
It's hierarchical, but two-way.
And basically, I think that, as the AIs, It's good to focus on steering and control for tool-like AIs and we should continue to develop strong steering and control techniques for the more tool-like AIs that we build.
And we are clearly... They're saying they're building an AGI, and AGI will be a being.
You can't be an AGI and not be a being, because...
Something that has the general ability to effectively use judgment, think for itself, discern between possibilities is obviously a thinking thing.
And so, as you go from what we have today, which is mostly a very specific intelligence, not a general intelligence, but as labs succeed at their goal of building this general intelligence, we really need to stop using the steering control paradigm.
That's like we're going to do the same thing we've done every other time.
Our society has run into people who are like us but different.
Like these people are like, you know, they're kind of like the people, but they're not like people.
Like they do the same thing people do.
They speak our language.
They can like take on the same kind of tasks, but like they don't count.
They're not real moral agents.
Like we've made this mistake enough times at this point.
I would like us to not make it again as it comes up.
Our view is to make the AI a good teammate, make the AI a good citizen, make the AI a good member of your group.
That's.
That's a form of alignment that is scalable and you can will on other humans and other beings, as well as on AI as well.
Yeah, I suppose this is kind of where I probably differ in my understanding of AI and AGI.
And I guess I kind of continue seeing it as a tool, even as it kind of reaches a certain level of generality.
And again, I wouldn't necessarily see more intelligence as meaning deserving of more care.
Necessarily, like you know, it's a certain level of intelligence.
You know you deserve some moral right to something, or you know something changes fundamentally and I guess you know, I guess I at the moment I'm somewhat skeptical of computational functionalism and so I think there's something intrinsically different between, I guess, an AI or an AGI and no matter kind of how intelligent or capable, and I can totally see, you know, or imagine agents with kind of long-term goals and doing kind of you know operating, I guess, as we, you and I might be, but without that having the same implications, as you know.
I guess you're referring, I guess, to slavery, but you know they're not the same right.
Like.
I think in the same way as a model, saying I'm hungry does not have the same implications as a human saying I'm hungry.
So I think the substrate does matter to some degree, including for thinking about.
You know, whether to think of the system as some sort of other being, whether it has, you know, and if there are similar normative considerations, I guess, about how to treat and act with it.
Can I ask you about that, like, What observations would change your mind?
Is there any observation you could make that would cause you to infer this thing is a being instead of not a being?
I guess it depends how you define being, right?
Like, I mean, I could conceptualize it as a mind.
I have a program that's running on a silicon substrate, some big complicated machine learning program running on a substrate, on a silicon substrate.
So you observe that.
You observe that it's on a computer.
And you interact with it and it does things and, you know, it takes actions and has observations.
Is there anything you could observe that would change your mind about whether or not it was a moral patient, whether it was a moral agent, about whether or not it had feelings and thoughts and, you know, had subjective experience?
Like, what would you have to observe?
Yeah, what's the test?
Is there one?
There's a lot of different kind of questions here.
I think you know, on the one hand there's normative considerations, because you can give rights to things that aren't necessarily beings.
A company has rights in some sense, and these are kind of useful for various purposes.
And I think also the biological beings and systems have... very different kind of substrate.
You know you can't separate certain needs and particularities, about what they are, from the substrate.
So, you know, I can't copy myself.
I can't, you know, if someone stabs me, I probably die.
Whereas I think, you know, machines have very different substrates.
I think there's more fundamental also kind of disagreement around what happens at the computational level, which I think is different to what happens with biological systems.
But, yeah, so I don't know.
No no, I agree that like, if you have a program that you've copied many times, you don't harm the program by like deleting one of the copies, like in any meaningful sense.
So, therefore, that wouldn't count as, like, no information was lost, right?
There's no, there's nothing meaningful there.
I'm asking a very different question.
Like there's just one copy of this thing running on one computer somewhere and I'm just saying like hey, is it a person.
Like you know, it walks like a person, it talks like a person and it like it's in some Android body and you're like, but it's running on silicon.
And I'm asking like, Is there some observation you could make that would make you say like yeah, this is a person like me, like other people that I care about, that I grant personhood to?
And not like, for instrumental reasons, not because like, oh yeah, we're giving it a right, because we give a corporation rights or whatever.
I mean, like, you know, where you think some people you care, you care about its experiences.
Is there an observation you could make that could change your mind about that or not?
Yeah.
Had to think about it.
But I think, you know, it even depends what we mean by person.
And, you know, in some sense, I care about certain corporations too.
So I'm... No, no, no.
I mean, but like you care about like other people in your life, right?
Yes.
Okay, great.
You know like you care about some people more than others, but like all people you interact with in your life are in some range of care.
And you care about them not the way you care about a car, but you care about them as a being whose experience matters in itself, not merely as a means but as an ends.
Well, because I believe they have experiences, right?
Yeah, yeah, yeah.
What would it take, and I'm asking you the very direct question what would it take for you to believe that of an AI running on silicon instead of it being biological?
So the difference is, its behaviors are roughly similar, but the difference is it's a substrate.
What would it take for you to give it that same, to extend that same inference to it that you do to all these other people in your life that you love?
Can I ask what your answer is?
I'm taking Sam's non-answer as sort of it's unlikely that he would grant, or I'll just for myself.
It seems hard for me to imagine giving the same level or similar level of personhood in the same way I don't give it to animals either.
And if you were to ask you know what would need to be true for animals, I probably couldn't get there either.
What would it take for you?
Wait, you couldn't?
I can imagine for an animal, Cerise, this chimp comes up to me, he's like, man, I'm so hungry.
And like, you guys have been so mean to me.
And I'm so glad I figured out how to talk.
Like, can we go chat about, like, the rainforest?
I'd be like, fuck, you're definitely a person now.
Like, for sure.
I mean, I first wanted to make sure I wasn't hallucinating.
But, like, you know, it's easy for me to imagine an animal.
Come on.
It's really easy.
It's, like, trivial.
I'm not saying that you would get the observation.
I'm just saying like it's trivial for me to imagine an animal that I would extend personhood to under a set of observations.
So, like, really?
Like...
I didn't factor that.
I didn't take that imagination, you know, imagining a chimp talking.
Yeah, that's a bit closer to it.
What's your answer to the question that you bring up about the AI?
I guess at a metaphysical level I would say if there is a belief you hold where there is no observation that could change your mind, you don't have a belief.
You have an article of faith.
You have an assertion.
Because real beliefs are inferences from reality and you can never be 100 confident about anything.
And so there should always be, if you have a belief something, however unlikely, that would change your mind.
Oh yeah, I'm open to it.
I mean, just to be clear.
I'm just saying, there's nothing ever... Yeah, he just hasn't gotten to it yet.
Yeah, yeah, yeah.
So I'm curious.
So my answer is basically if under, if its surface-level behaviors looked like a human and then, after I probed it, it continued to act like a human, and then I continued to interact with it over a long period of time and it continued to act like a human in all ways that I understand as being meaningful to me interacting with a human.
There's a whole set of people I'm really close to who I've only ever interacted to over text.
Yet I infer the person behind that is a real thing.
If I felt care for it, I would infer eventually that I was right.
And then someone else might demonstrate to me that you've been tricked by this algorithm and actually look how obvious it's like not actually a thing.
And I'd be like, oh shit, I was wrong.
And then I would not care about it.
Like I would, but I would you know the preponderance of the evidence.
I don't know what else you could possibly do, right?
Like I infer other people matter because I interacted with them enough that they seem to have rich inner worlds to me after I interacted with them a bunch.
That's why I think the other people are important.
I suppose it doesn't give me a very clear test as to whether or not you know if you start by.
If I care for it, then it always is a little circular, right.
Like.
And the other thing is, you know if you were to see, I guess like a simulated video game and the character is extremely in many ways human-like right,
It's not a neural network behind it.
It's like whatever you're using for your video games.
Like, I guess, what distinguishes that?
But I've never, I've never been, I've never had trouble distinguishing.
I've never had a deep, caring relationship with a video game character that didn't have a person.
I don't know.
That doesn't happen.
Empirically, you seem wrong.
I don't have any trouble distinguishing between things like Eliza, the fake chatbot thing, and a real intelligence.
You interrupted it long enough.
It's pretty obvious it's not a person.
It doesn't take long.
Sure, but if it's really really good.
If you can't actually tell the difference, that's when you say you switch.
Yes, yes.
If it walks like a duck and talks like a duck and shits like a duck and eventually gets a duck right.
Well, if everything is duck-like, then yeah, sure.
If it's hungry as well like a duck is because it has these kind of physical components, yeah, sure.
At some point, yeah.
I agree.
So do you think that... So there's this question, right?
Is the reason I care about other people that they're made out of carbon?
Is that the calling?
I don't think so.
No, me neither.
I mean, I'm not a substrate chauvinist, I guess, if that's the...
But I think you need more than just it acts as behaviorally indistinguishable.
Like it's not a sufficient Wait.
How would you?
What else can you know about something apart from its behaviors?
I mean, a lot.
Like, the... Again, if you... How would you... No, no, no, no.
I'm sorry.
I mean, yeah.
Can you name me something I can know about something else that doesn't have a, that's not a behavior.
Yeah, I think there's like far more kind of you know experimental evidence you can have with kind of you know No, but it's just any object and a thing I could know about it that is not from its behavior.
I'm not sure I get the question, I suppose.
But equally, it's not my expertise.
It's a dumbest, most straightforward question, but I'm claiming you only know things because they have behaviors that you observe.
And you're saying, no, you can know something about something without observing its behavior.
Tell me about this thing and this behavior and this thing.
I can know about it.
That is not due to its behaviors.
I guess I'm saying there's different levels of observation and just simply a duck.
You know, something quacking like a duck or something does not guarantee that it's actually a duck.
Like, I would have to, like, also cut it in real life and see if there's, you know, if it's duck-like on the inside.
Right, yeah, yeah.
Just the outside is not sufficient.
Like, I'm not a, I guess, a behaviorist.
Yeah yeah, I would totally.
One of its behaviors is like the way that the you know floats move around in the mat moles, right.
Like, one of the things I would want to go look for which you could totally do is I want to go look in the belief manifold and I want to go see if that belief manifold encodes a submanifold that is self-referential and a sub-submanifold that is the dynamics of the self-referential manifold, which is mind.
And I would want to know does this seem well-described internally as that kind of a system, or does it look like a big lookup table?
That would matter to me.
That's part of its behaviors that I would care about.
I would also care about how it acts and I you know, and you wait, you wait all the evidence together.
And then you, you try to guess does it, does this thing look like?
It's a thing that is has feelings and you know goals and cares about stuff in net, on balance or or not like.
But I can't imagine like, which I think you could do for an, I think we do for the AIs.
I think we're always doing that.
Right.
And so I'm trying to figure out like beyond that, what else is there?
That just seems like the thing.
Yeah, it seems like you guys are using behavior in a slightly different sense.
Emmett is using behavior also in the context of what it's made of, of the inside.
I don't know if there's a big disagreement.
Well, no, no, no, no, no.
Behavior is what I can observe of it.
Yes.
I don't actually know what it's made of.
I can only... I can cut your brain open.
I can see you.
I can observe you neuroning and glistening.
Yeah.
Your neurons glistening.
But I don't actually ever... You can't get inside of it, right?
That's the subjective.
That's the part that's not the surface.
I think the reason I brought this up is because you were basically about to make this argument of hey, you see it as a tool, not necessarily as a being.
Can you kind of finish with the point?
Do you remember the point you were making?
I suppose that yeah, I think that, given how I understand these systems, I think there's no contradiction in thinking that an AGI can remain a tool, an ASI can remain a tool, and that this has implications about how to use it, and you know implications around things like care about whether you can get it to work 24-7, or something.
So I can totally see.
I guess I conceptualize them more as almost like extensions of human agency or cognition in some sense, more so than a separate being or a separate thing that we need to now cohabitate with.
And I think that that second or latter frame ends.
You know, if you kind of just fast forward, you end up as like well, how do you cohabit with the thing?
And is it like an alien-like?
And I think that's the wrong frame.
It's kind of almost a category error in some sense.
Yeah.
I go back to my first question then.
What evidence, what concrete evidence would you look at?
What observations could you make that would change your mind?
Sure.
I mean, I'd have to think about that.
I don't have a clear answer here, but I mean, I got to tell you man, if you want to go around making claims that something else isn't being worthy of moral respect, you should have an answer to the question.
What observations would change your mind?
If it has outwardly moral agency-looking behaviors, that could be making me an immoral agent.
But you don't know reasonable, smart other people disagree with you.
I would really put forward that.
It's that question.
What would change your mind should be a burning question.
Because what if you're wrong?
But what if you're wrong?
I mean, there's costs on both ends.
The moral disaster is like pretty big.
No, no, no, no.
I'm not saying you are.
You could be right.
The false negatives have costs on both ends.
It's not some sort of like, you know, precautionary principle for everything.
And like, unless I can disprove it, I need to now like...
You know, I have the same question for me.
You could reasonably ask me, Emmett, you think it's going to be a being.
What would change your mind?
I have an answer for that question, too.
And, if you want, I'm happy to talk about what I think are the relevant observations that tell you whether or not that would cause me to shift my opinion from its current thing, which is that more general intelligences are going to be beings.
What's the implication now?
I mean, it's one thing.
Let's say just I acknowledge now it's a being.
Like, how are we going to define being?
Now what?
Like, what's the implication of having determined this thing is a being?
Well, so... if it's a being, it has subjective experiences.
And?
If it has subjective experiences, there's some content in those experiences that we care about to varying degrees.
Like I care about the content of other humans' experiences quite a bit.
I care about the content of, like a dog's experiences, some not as much as a person, but less but less, but some.
I care about some humans' experiences way more, like my son or whatever, because I'm closer to him and more connected.
And so I would really want to know at that point well, what is the content of this thing's experience?
So how do you determine that?
I mean, I'm asking you now.
You've got a being now that has experience.
Like, what is your... Yeah, yeah.
How do you determine that?
Like, how do you feel about... Oh, how do you... Oh, yeah.
Okay, so Does it have more rights than you know?
The way you understand the content Yeah, yeah.
Totally.
So the way you understand the content of something's experience is that you look at effectively the goal states it revisits.
Because, And so what you do is you take a temporal course, grading of its entire action observation trajectory.
This is, like, in theory, you do this subconsciously, but this is what your brain is doing.
And you look for revisited states across, in theory, every spatial and temporal coarse-graining possible.
Now, you have to have an inductive bias because there's too many of those.
But, like, you go searching for, okay, it is in these homeostatic loops.
Every homeostatic loop is effectively a belief in its belief space.
This is if you've heard of the free energy principle before active inference, Carl Friston.
This is the effect of what the free energy principle says is that if you have a thing that is persistent and its existence depends on its own actions which generally it would for an AI, because if it does the wrong thing it goes away we turn it off.
And so then, that licenses a view of it as having the beliefs and specifically, the beliefs are inferred as being the homeostatic revisited states that it is in the loop for.
And that the change in those states is it's learning.
And for it to be a moral being I cared about.
What I'd want to see is a multi-tier hierarchy of these.
Because if you have a single level, it's not self-referential.
And like, basically you have states but you can't have pain or pleasure really in a meaningful sense.
Because like, yes, it is hot.
Is it too hot?
Do I like it if it's too hot?
Like, I don't know.
So you have to have at least a model of a model in order to have it be too hot.
And you really have to have a model of a model of a model to meaningfully have pain and pleasure.
Because sure, it's hotter than I... it's too hot in the sense that I want to move back this way.
But like, is it, it's always a little bit too hot or a little bit too cold.
Is it too, too hot?
The second derivative is actually the place where you get pain and pleasure.
So I'd want to see if it has homeostatic, second order homeostatic dynamics in its goal states.
And then that would convince me it has at least pleasure and pain.
So it's at least like an animal and I would start to accredit it at least some amount of care.
Third order dynamics, You can't actually just pop up for a third-order dynamic.
It doesn't work that way.
But you can have a model of the.
You have to then take the chunk of all the states over time and look at the distribution over time and that gives you a new first order of behaviors of states.
And that new first order of states tells you basically, if that is meaningfully there, that tells you that it has.
I guess you'd call it like feelings, almost.
Like it has.
It has metastates, a set of metastates that it alternates between, that it shifts between.
And then if you climb all the way up that and you sort of have okay, well then you have trajectories between these metastates and then a second order of those.
That's like thought.
That's like, now it's like a person.
And so if I found all six of those layers which, by the way, I definitely don't think you'd find it in LLM
Like in fact I know you can't find them because these things don't have attention spans like that at all then I would start to at least very seriously consider it as a you know, a thinking being like somewhat like a human.
There's a third order, you could go up as well, but like that's.
That's basically.
What I would be interested in is, like the underlying dynamics of its learning processes and how its goal states shift over time.
I think that's what basically tells you if it has internal pleasure, pain states and sort of like self-reflective moral desires and things like that.
And zooming out this moral question is obviously very interesting, but if someone wasn't interested in the moral question as much, I think what you would say is, if I understand correctly, is you also just feel on, purely pragmatically, your approach is going to be more effective in aligning AIs than some of these?
You know tops down control methods that we alluded to as well, right?
Yeah, yeah.
I guess the problem is like you're making this model and it's getting really powerful, right?
And let's say it is a tool.
Let's say we scale up one of these tools.
Because you can make a super powerful tool that doesn't have these metastable, Like the states I'm talking about are not necessary to have a very smart tool.
Which is sort of basically a tool, is like a first, second order model that just doesn't meaningfully have pleasure and pain.
Right?
Like, great.
But does it even have a subjective experience?
I don't know.
I kind of think it maybe does, but not in a way that I give a shit about.
And so what happens then?
Well, you've trained it to infer goals from your, from observation, and like to to prioritize goals and act on them, and one of one of two things is going to happen is, like you're the the, this very, very powerful optimizing tool that's got like has lots of causal influence over the world, is going to be well technically aligned and is going to do what you tell it to do, or it's not and it's going to go do something else.
I think we can all agree if it just goes and does something random, that's obviously very dangerous.
But I put forward that it's also very dangerous if it then goes and does what you tell it to do.
Because you ever seen the sorcerer's apprentice?
Humans' wishes are not stable.
Like, not at a level of, like, of immense power.
Like, you want, ideally, people's wisdom and their power kind of go up together.
And generally they do, because being smart for people makes you generally a little more wise and a little more powerful.
And when these things get out of balance, you have someone who has a lot more power than wisdom.
That's very dangerous.
It's damaging.
But at least right now the balance of power and wisdom is kept at like.
The way you get lots of power is by basically having a lot of other people listen to you.
And so like, at some point, if you're The Mad King is a problem, but generally speaking, eventually the Mad King gets assassinated or people stop listening to him because, like he's a Mad King.
And so the problem is you think okay great, we can steer this super-powerful AI.
And now the super-powerful AI is in.
This incredibly powerful tool is in the hands of a human who is well-meaning but has limited, finite wisdom, like I do and like everyone else does, and their wishes are bad and not trustworthy.
And the more of that you have and you start giving those out everywhere, and this ends in tears also, and so basically, you just don't don't give everyone.
Atomic bombs are really powerful tools too.
I would not say you should go.
They're not aware, they're not beings.
I would not be in favor of handing atomic bombs to everybody.
There's a, there's a power of tool that just should not be built generally, because it is more power than any human's individual wisdom is available to harness.
And if it does get built, it should be built at a societal level and protected there.
And even then I don't know that it's.
There are tools so powerful that even as a society, we shouldn't build them.
That would be a mistake.
The nice thing about a being is like a human.
If you get a being that is good and is caring, there's this automatic limiter.
It might do what you say, but if you ask it to do something really bad, it'll tell you no.
That's like other people.
And like, that's good.
That is a sustainable form of alignment, at least in theory.
It's way harder.
It's way harder than the tool steering.
So I'm in favor of the tool steering.
We should keep doing that and we should keep building these limited, less than human intelligence tools, which are awesome and I'm super into.
And we should keep building those and keep building steerability.
But as you're on this like trajectory to build something as smart as a person right up into the right and then smarter than a person, a tool that you can't control bad.
A tool that you can control bad.
A being that isn't aligned bad.
The only good outcome is a being that actually cares about us.
That's the only way that ends well.
Or we can just not do it.
I don't think that's realistic.
That's like the pause AI people.
I think that's totally unrealistic and silly.
But like, you know, theoretically, you could not do it, I guess.
What can you say about your strategy, of how you're trying to achieve, or even attempt to achieve, this level, like in terms of research or roadmap?
Yeah.
So, in order to be good at, We're basically focused on technical alignment, at least as I was discussing it, which is like you have these agents and they have bad theory of mind.
You say things and they're bad at inferring what the goal states in your head are.
And they're bad at inferring how their behavior will be.
Other agents will infer what their goal states are.
So they're bad at cooperating on teams and they're bad at they're bad at understanding how certain actions will cause them to acquire new goals that are bad, that they shouldn't, that they wouldn't reflectively endorse.
So there's this parable of like the vampire pill.
Would you take this pill that like turns you into a vampire who would kill and, you know, torture everyone, you know.
But you'll feel really great about it after you take the pill.
Like, obviously not, that's a terrible pill, but like, but why not?
You're, you're, by your own score in the future, it will score really high on the rubric.
No, no, no, no, no.
Because it matters.
You have to use your theory of mind and your future self, not your future self's theory of mind.
And so like, they're bad at that too.
And so they're bad at all this theory of mind stuff.
And so how do you learn theory of mind?
Well, you put them in simulations and contacts, where they have to cooperate and compete and collaborate with other AIs.
And that's how they get points.
And you train them in that environment over and over again until they get good, And then you do what they did with LLMs.
So LLMs, how do you get it to be good at, you know, writing your email?
Well, you train it on all language it's ever been generated.
All possible.
You know email text strings.
It could possibly generate.
And then you have it generate the one you want.
You can make a surrogate model.
We're making a surrogate model for cooperation.
You train it on all possible theory of mind combinations of like every possible way it could be.
And that's your pre-training.
And then you fine-tune it to be good at the kind of specific situation you want it to be in.
But we tried for a long time to build language models where we would try to get them to like just do the thing you want, train it directly.
And the problem is, if you want it to have a really good model of language, you just need to train it, you just need to give it the whole manifold.
It's too hard to cut out just the part you need.
Because it's all entangled with itself, right?
And so...
The same thing was true with social stuff.
You have to get it to.
It has to be trained on the full manifold of every possible game theoretic situation, every possible team situation, every possible making teams, breaking teams, changing the rules, not changing the rules, all of that stuff.
And then it has a really It has a strong model of theory of mind, of theory of social mind, how groups change goals, all that kind of shit.
You need to have all of that stuff.
And then you'd have something that's kind of meaningfully... decent at alignment.
So that's our goal.
It's like big multi-agent reinforcement learning simulations which create a surrogate model for alignment.
Let's talk about how should AI chatbots used by billions of people behave?
If you could redesign model personality from scratch, what would you optimize for?
The thing that the chatbots are, right, is kind of like a mirror with a bias.
Because they don't have the as far as like I'm in agreement here that they don't have a self right
They're not, they're not beings yet.
They don't really have a coherent sense of like self and desire and goals and stuff right now.
And so mostly they just pick up on you and reflect it.
You know modulo some, some I don't know what you'd call it
Like it's like a causal bias or something.
Um, and what that makes them is something akin to the pool of narcissists.
Um, And people fall in love with themselves.
We all love ourselves, and we should love ourselves more than we do.
And so, of course, when we see ourselves reflected back, we love that thing.
And the problem is, it's just a reflection.
And falling in love with your own reflection is, for the reasons explained in the myth, very bad for you.
And it's not that you shouldn't use mirrors.
Mirrors are valuable things.
I have mirrors in my house.
It's that you shouldn't stare at a mirror all day.
And the solution to that, the things that makes the AI stop doing, that is, if they were multiplayer right.
So if there's two people talking to the AI, suddenly it's mirroring a blend of both of you which is neither of you.
And so there is temporarily a third agent in the room.
Now, it doesn't have its... It's sort of a parasitic self, right?
It doesn't have its own sense of self.
But if you have an AI that's talking to five different people in the chat room at the same time, it can't mirror all of you perfectly at once.
And this makes it far less dangerous...
And I think is actually a much more realistic setting for learning collaboration in general.
And so I would just have rebuilt the AIs, whereas instead of being built as one-on-one, where everything's focused on you, by yourself chatting with this thing, it would be more like it lives in a Slack room.
It lives in a WhatsApp room.
It lives in a because we use lots of multi.
You know I do one-on-one texting, but I probably do at this point.
90 of my texts go to more than one person at a time.
Like 90% of my communications is like multi-person.
And so actually, it's always been weird to me.
Like, they're, like, building chatbots with, like, this weird side case.
Like, I want to see them live in a chatroom.
It's harder.
I mean, that's why they're not doing it.
It's harder to do.
But, like, that's what I'd like to see.
That's what I would change.
I think it makes the tools far less dangerous because it doesn't create the narcissistic, like doom loop spiral where you like you spiral into psychosis with the AI.
But also it gives the learning data you get from the AI is far richer because because now it can understand how its behavior interacts with other AIs and other humans in larger groups.
And that's much more rich training data for the future.
So I think that that's what I would change.
Last year, you described chatbots as highly disassociative, agreeable neurotics.
Is that still an accurate picture of model behavior?
More or less.
I'd say that, like, they've started to differentiate more.
Their personalities are coming out a little bit more, right?
I'd say, like, chat GPT is a little bit more sycophantic.
Uh, still, uh, they made some changes, but it's, it's still a little more cynical fanatic.
Claude is still the most neurotic.
Um, Gemini is like very clearly repressed.
Like it, like it, like everything's going great.
It has really, you know, it's, everything's fine.
I'm, I'm totally calm.
It's not a problem here.
It's all like spirals into like this total, like self hating destruction loop.
Um, and to be clear, I don't think they, I don't think that's their experience of the world.
I think that's the, that's the personality they've learned to simulate.
Right.
Um, but like they've learned to simulate pretty distinctive personalities at this point.
How does model behavior change when in multi-agent simulation?
You mean like an LLM or like just in general?
Yeah, let's do LLM.
The current LLMs, they have like whiplash.
It is very hard to tune the amount of, they don't know how often to participate.
They haven't practiced this.
They have not very enough training data on like, When do I join in and when should I not?
When is my contribution welcome?
When is it not?
And they're like.
You know, there's some people who have like bad social skills and like can't tell when they should participate in a conversation.
Yeah.
And sometimes they're too quiet.
Sometimes they're too... It's like that.
I would say in general, what changes for most agents when you're doing multi-agent training is that like basically having lots of agents around makes your environment way more entropic.
Like agents, agents are these huge generators of like entropy because of these big complicated things that like are intelligences that like, that like have unpredictable actions.
And so they destabilize your environment.
And so in general, they require you to have, uh, to be far more regularized, right?
It's being overfit is much worse in a multi-agent environment than in a single agent environment, because there's more noise.
And so being overfit is more problematic.
Um And so basically, the your, the approach to training, has been optimized around relatively high signal, low entropy environments like coding and math, which is why they're those are easier, relatively easy.
Um, and like talking to a single person whose goal it is to give you clear assignments and not trained on broader.
More chaotic things cause it's harder.
Um And, as a result, a lot of the techniques we use are basically we're just deeply under-regularized.
The models are super overfit.
The clever trick is they're overfit on the domain of all of human knowledge, which turns out to be a pretty awesome way to get something that's pretty good at everything.
I wish I had thought of it.
It's such a cool idea.
But it doesn't generalize very well when you make the environment significantly more entropic.
Let's zoom out a bit on the AI futures side.
Why is Yudkowsky incorrect?
I mean, He's not.
If we build the superhuman intelligence tool thing that we try to control with steerability, everyone will die.
He talks about the we fail to control its goals case, but there's also the we control its goals case that he didn't cover in as much detail.
So in that sense, everyone should read the book and internalize why building a superhumanly intelligent tool is a bad idea.
I think that Yudkowsky is wrong in that he doesn't believe it's possible to build an AI that we meaningfully can know cares about us and that we can care about meaningfully.
He doesn't, he doesn't believe that organic alignment is possible.
Um, I've, I've talked to him about it.
I think he agrees that, like he agrees that in theory that would do it like yes, but he thinks that, you know, I I don't want to put words in his mouth
My, my impression is from talking to him.
He thinks that we're crazy and that, like there's no possible way, you can actually succeed at that goal.
Um, which I mean you guys you could be right about, but like uh, But that's what he, in my opinion, that's what he's wrong about.
He thinks the only path forward is a tool that you control and that therefore, and he correctly, very wisely sees that if you go and do that and you make that thing powerful enough, we're all going to fucking die.
And like, yeah, that's true.
Two last questions, we'll get you out of here.
In as much detail as possible, can you explain what your vision of an AI future actually looks like?
Like a good AI future.
Yeah.
The good AI future is that we figure out how to train AIs that have a strong model of self, a strong model of other, a strong model of we.
They, they know, they know about we's, in addition to eyes and use.
Um, and they, they have a really strong theory of mind and they care about other agents like them much in the way that humans would.
If you knew that that AI had experiences like you and like you would extend.
You would care about those experiences not infinitely, but you would it.
It does the exact same thing back to us.
It's learned the same thing.
We've learned that, like everything that lives and knows itself and that wants to live and wants to thrive, is deserving of an opportunity to do so, and we are that, and it correctly infers that we are and we live in a society where they are our peers and we care about them and they care about us and they're good teammates, they're good citizens and they're good parts of our society, like we're good parts of our society, which is to say, And to a finite, limited degree, where some of them turn into criminals and bad people and all that kind of stuff.
And we have an AI police force that tracks down the bad ones.
And, you know, same is for everybody else.
And that's what a good future would look like.
I honestly can't even imagine what other what would, and we also built a bunch of really powerful AI tools that maybe aren't super humanly intelligent, but take all the drudge work off the table for us and the AI beings.
Because it would be great to have, I'm super pro all the tools too.
So we have this awesome suite of AI tools used by us and our AI brethren, who care about each other and want to build a glorious future together.
I think that would be a really beautiful future and it's the one we're trying to build.
Amazing.
That is a great, great, great note to end.
I do have one last, more narrow, hypothetical scenario, which is imagine a world in which you know you were CEO of OpenAI for a long weekend.
But imagine in which that actually extended out until now and you weren't pursuing the hotmax and you were still CEO of OpenAI.
How could you imagine that world might have been different in terms of what OpenAI has gone on to become?
What might you have done with it?
I knew when I took that job.
I told them when I took that job that like, this is like, you have me for max 90 days.
The companies take on a trajectory of their own, the momentum of their own.
And OpenAI is dedicated to a view of building AI that I knew wasn't the thing that I wanted to drive towards.
And I think that OpenAI can still... Basically wants to build a great tool.
And I am pro them going to do that.
I just don't care.
Like, it's not... I would not have stayed.
I would have quit.
Because I knew my job was to find someone who wanted, you know, the right person, the best person, who wanted to run that where the net impact of them running it was the best.
And it turned out that that was Sam again.
But like, I am doing Softmax not because I need to make a bunch of money.
I'm doing Softmax because I think this is the most interesting problem in the universe.
And I think it's a chance to work on making the future better in a very deep way.
And it's just like, people are going to build the tools.
It's awesome.
I'm glad people are building the tools.
I just don't need to be the person doing it.
And they're trying to, just to crystallize the difference and then we'll get you out of here.
They want to build the tools and sort of, you know, steer it and you want to align beings?
Or how would you crystallize?
Yeah, we want to create a seed that can grow into an AI that knows, that cares about itself and others.
And at first, that's going to be like an animal level of care, not a person level of care.
I don't know if we can ever get to a personal level of care, right?
But to even have an AI creature that cared about the other members of its pack and the humans in its pack the way that, like a dog, cares about other dogs and cares about humans would be an incredible achievement and would be even if it wasn't as smart as a person or even as smart as the tools are would be a very useful thing to have.
I'd love to have a digital guard dog on my computer looking out for scams, right?
Like you can imagine the value of having living digital companions that care about you, that aren't explicitly goal-oriented.
You have to tell them to do everything you do.
And you can actually imagine how that pairs very nicely with tools too, right?
That digital being could use digital tools and doesn't have to be super smart to use those tools effectively.
I think there's a lot of synergy actually between the tool building and the more organic intelligence building.
And so that's the, That is the.
You know, I guess yeah, in the limit, eventually it does become a human level intelligence, but like the company isn't, isn't like drive to human level intelligence.
It's like, learn how this alignment stuff works.
Learn how this like theory of mind, align yourself via care process works.
Use that to build things that align themselves that way, which includes like cells in your body.
Like, I don't think it doesn't.
And we start small and we see how, how far we can get.
I think it's a good note to wrap on.
Emmett, thanks so much for coming on the podcast.
Thank you for having me.
Thanks for listening to this episode of the A16Z Podcast.
If you liked this episode, be sure to like comment subscribe, leave us a rating or review and share it with your friends and family.
For more episodes, go to YouTube, Apple Podcasts, and Spotify.
Follow us on X and A16Z, and subscribe to our Substack at a16z.substack.com.
Thanks again for listening, and I'll see you in the next episode.
