I think a major obstacle is a lack of trust in the government in general. Almost nobody expects the government to do right by its constituents anymore, even when it would cost very little. Removing an entire industry - and an industry that appears to be the only healthy/growing segment of the economy, at that - seems beyond the pale.
While rhetoric differs, there seems to be a bipartisan acceptance that, if the people currently in power wish to remain so, they'll need a wunderwaffen. The military suffers from a recruiting and innovation crisis, the education system seems to have been co-opted by rent-seekers even as every success metric careens downwards, manufacturing has the occasional success story but remains largely stagnant, and the only point of unity between the two parties' voter bases is that the status quo is intolerably bad, perhaps to the point of necessitating a civil conflict.
AI may not be popular, but the people who believe it will fix everything appear to be the only group of people who think anything is going to be fixed, and the people in charge are inclined to listen to the one group telling them that there's a chance for them to recover their former prestige. The choice that seems to be before them is "gamble on AI, and maybe win power for the rest of history" versus "don't gamble, and definitely lose, either to China, or to populist reform, or to a gradual decay into irrelevance". At the very least, I think that's how the segment of the public concerned about AI views the elite's decision-making process.
1This is a reason for "stop" over "regulate;" it takes less dexterity to wield a hammer than scalpel.
While i agree with this post almost entirely, i can't pretend that i still feel like we have no choice. Trying to stop the development of AI today, is kind of like trying to have stopped the manhattan project. Even worse in my opinion if i think about it: the profit incentive is there, the military incentive is there. And from history we've seen that public disapproval doesn't necessarily stop governments or corporations from anything, if the incentive is large enough. The whole world is already essentially operating as a large machine with only one goal: maximize profits. And it does so no matter the cost, which is why we have more slaves today than ever before in history. All that being said, it can't hurt to try our best. But i also cannot pretend the feeling of hopelessness is not there, and that it's not (at least to my current understanding) well justified. Whenever we think of cases where something changed because of public disapproval, it was likely because the consequences of that disapproval outweighed the benefits of continuing. But with AI, that's simply not the case.
1The mechanism isn't [public disapproval] -> [AI moratorium]. That sounds hopeless to me, too. What I'm actually relatively optimistic about is [public awareness] -> [educating policymakers] + [signal to policymakers that they are allowed to do something about it] -> [common knowledge of the desire to act] -> [AI moratorium]. This feels very doable to me after getting back from speaking with my congressional offices in DC.
The key here is that most politicians don't want to take these risks either. Really very few people are on board with the whole "create superintelligence we don't know how to control or make care about us" thing. If enough politicians and policymakers learn that the "create superintelligence" thing necessarily brings a risk of human extinction, one would hope that they would act out of their own fear and desire for self-preservation. But because of how politics works (or doesn't), especially in the US, it takes a lot to overcome the strong social fear of speaking out. They need to know that their constituents have their backs, as opposed to either supporting the status quo or having very different priorities.
So public disapproval isn't a lever with which to move the whole world. It's just a firm tap to send the balance in the direction of action. The whole mechanism is being built simultaneously in parallel, and the education and advocacy work gets easier the more the public is engaged with the issue. The bigger the tap, the smaller the gap.
Trying to stop the development of AI today, is kind of like trying to have stopped the manhattan project.
You are correct that stopping the Manhatten project would have been much easier.
It was a secret project, and the incentives for most of the decision-makers involved were around producing "better-but-accountable" outcomes. There were multiple individuals to whom if you presented a sufficiently plausible argument, might have been able to stop the Manhattan project, up until the actual attack on Hiroshima. Fermi et al. had a pretty good chance of preventing the bomb from being used, and had other technical experts agreed that the risks were too high, I think the Manhattan project could have been stopped.
The current situation is much worse, with far more people having strong incentives to deceive others, themselves, and knowingly take risks.
All of that said, I still don't think this means we have no choice. There are multiple ways of coordinating to solve a bad equilibrium, and I don't think the incentives for pursuing AI are nearly so universal.
Most of the public recognizes that they have a lot of incentive to stop the development of ASI, even if right now it is more focused on job losses, they also recognize that the gains of AI are unlikely to be distributed evenly, and most Americans in particular care more about their chance of gaining in social status, than almost any other terminal outcome of policy (I can provide some citations if needed.)
More importantly, that public disapproval likely has a stable equilibrium in that they probably should be worried about the existential risk, whatever percentage it really is.
All that needs to follow is that the public disapproval needs to be strong enough and translate into a strong enough signal that policy makers are duly incentivize to take the concerns seriously, and policy make.
That isn't super easy, but it isn't pure fantasy at all. Everyone can see how the incentives for AI firms favor total race dynamics, but for most of humanity and even most Americans incentives are more than strong enough to prefer caution.
Possibly a hot take here, but I don't think collective public opinion can reach a scale to prevent society from producing an AI capable of causing massive damage. Just because we all hate something, doesn't make that thing go away, even if it's something which appears to be under our control. Most people hate going to work Monday morning, and technically we could all choose not to, but the larger superstructure forces us to get out of bed.
Dangerous AI development may be similar, it may be an inevitable byproduct of our society existing in the first place. The incentive to be the first to create it first is very, very high, with AI companies already producing (though not publishing) AI which can be used for malicious purposes, i.e. Mythos. The production capability is there, as is the constant competition between AI companies which pushes them to continue development, even if it's not in the collective self-interest.
A common thought pattern people seem to fall into when thinking about AI x-risk is approaching the problem as if the risk isn’t real, substantial, and imminent even if they think it is. When thinking this way, it becomes impossible to imagine the natural responses of people to the horror of what is happening with AI.
This sort of thinking might lead one to view a policy like getting rid of advanced AI chips is “too extreme” even though it’s clearly worth it to avoid (e.g.) a 10% chance of human extinction in the next 10 years. It might lead one to favor regulating AI, even though Stopping AI is easier than Regulating it. It might lead one to favor safer approaches to building AI that compromise a lot on competitiveness, out of concern that society will demand a substitute for the AI that they don’t get to have.
But in fact, I think there is likely a very narrow window between “society not being upset enough to do anything substantial to govern AI” and “society being so upset that getting rid of advanced AI chips is viewed as moderate”.
There are a few reasons why I think people are likely to favor stopping AI over other policies, once they are taking the problem seriously.
Concern about the other risks of AI
The discussion about AI is often framed as utopia or dystopia; see, e.g. “The AI Doc: Or How I Became an Apocaloptimist”. A lot of the people most concerned about human extinction from rogue AI think that, unless it kills us, AI is going to be great.
But I think for most people, “AI is going to be super powerful” is already enough cause for concern. “AI is going to take everyone’s jobs, but don’t worry the AI companies are in control of the AI” doesn’t sound very reassuring to most people. I think it’s pretty clear that society isn’t entirely ready for the massive changes that AI can bring. I’m not saying disaster is guaranteed, just that making a general-purpose replacement for humans is something we could use more time to prepare for. I think many people will feel this, intuitively. There are lots of things we can do to try and prepare -- that’s kind of the point of other policies -- but I think more piece-meal / band-aid type approaches will leave us struggling to keep up.
I think another deal-breaker for some proposals might be concerns about mass surveillance and concentration of power. If AI is extremely powerful, then many people will reject proposals that rely on giving governments or other bodies power over AI.
The KISS principle: Keep it Simple, Stupid
I think “Stop AI” is an extremely simple and intuitive idea that people can understand and trust. It seems hard to get the details right with other plans, and many of them seem to rely critically on the details. Get it wrong and you can end up with dangerous AI slipping through the cracks and causing catastrophes, or central authorities having too much power.
I think people will be scared of the immense power of AI, and will have a hard time trusting any system to govern that power. Technocratic solutions that require expert knowledge to understand will be especially hard for society to accept.
A preference for humans remaining relevant
I think a lot of people simply won’t like the idea of humans being rendered obsolete. The idea of such a world will make them uncomfortable and sad. It is incompatible with all of their aspirations, their vision of the future, and the life they hoped for for themselves, their loved ones, their community, and their country.
This could certainly change over time, but I think in the immediate term, a lot of people simply won’t want to give up humanity’s privileged place as the most intelligent species, and the one that matters. Without a significant slowdown, I think it would be difficult to find a way to integrate AI into society that doesn’t threaten all of this, and so people with such preferences will find this whole AI thing is too much, too fast.
Thanks for reading The Real AI! Subscribe for free to receive new posts and support my work.
Share