Socialism Has Never Been More Important Than in the Age of AI
An AI worrier's case for leftist politics
I have previously written about why I think AI might wipe out all of humanity in the not so distant future. I have also previously written about my support for a traditional left wing economic agenda, based squarely on egalitarian values. In this article, I will attempt a synthesis of these two topics. More specifically, I will argue that an egalitarian, left wing economic agenda has never been more important than it is now, in virtue of the threats and opportunities posed by AI.
I believe that socialist ideas are of massive importance in the era of AI for two central reasons:
The fact that AI development is occurring under capitalism substantially increases the probability that literally everybody on earth will die as a result of this development. Moving towards socialism, then, will help to ensure that the human species makes it into the future at all.
If we do manage to develop highly capable AI under capitalism without killing literally everybody, the resulting world will likely be one of massively heightened economic inequality and social domination. Moving towards socialism, then, will help to ensure that the benefits of AI are instead used to secure the flourishing of everyone.
In the sections that follow, I will elaborate on and defend these two points.
Competing for Extinction
Capitalism is an economic system which is based, fundamentally, on private actors blindly chasing profit while largely ignoring the consequences that their decisions impose on humanity as a whole. The capitalist’s prime directive can be captured by a very simple formula: capitalists want to incur as much revenue as possible while incurring as little cost as possible. This is, at the end of the day, all a private company really cares about. The problem, of course, is that many of the costs which are incurred by society or even humanity broadly as a result of the capitalists activity do not appear directly as costs on the balance sheet of the capitalist- that is, they do not directly and immediately impact the cash flow of the business. This dynamic causes well known problems.
For example, it has become overwhelmingly clear that in order to maintain a suitable environment for human life, we must move away from the usage of fossil fuels, and build out a clean energy infrastructure to replace them. However, since the production of fossil fuels continues to be profitable, and the building of a clean energy infrastructure to replace them continues to be not sufficiently profitable, we will continue to drift towards ecological destruction as long as the capitalist market is left to its own devices. As some of you might recall, it was not the market, left to its own devices, but rather government intervention, which ended the production of environmentally destructive chloroflourocarbons, and forced us to use other, less destructive chemicals in our hair spray. If the market had been left to its own devices, such chemicals would no doubt still be produced en masse.
On the flipside, we can think of various things which would be useful for humanity, but which aren’t provided by capitalists because they are not profitable to provide. For example, it would be largely beneficial from the standpoint of human well being to provide healthcare for everyone. However, because doing so would represent a big financial loss to insurance companies, we have millions of US citizens who go uninsured. Similarly, it would be very useful for internet companies to spread high speed internet to rural communities. However, since these communities are not sufficiently densely populated for such a move to be profitable, this is not done.
Importantly, if some capitalist does not behave in ways that are conducive to profit maximization, they will simply go out of business, as they will be outcompeted by another capitalist who does. It’s not just that capitalists want to maximize their profit- it’s that they need to, if they are to continue being capitalists. Maximizing ones profit is, then, a necessary condition for the survival of the capitalist in a market.
This dynamic- that is, the dynamic of blind, competitive profit maximization, is particularly dangerous in the context of AI. If this technology really is as dangerous as I’ve suggested, we probably don’t want a bunch of uncoordinated private firms, each of whom is single mindedly focused on profit, all racing to bring it into existence as quickly as possible with no public supervision and little regard for the possible consequences to humanity. Getting AI development right is one of the primary tasks, if not the primary task, of this century. Furthermore, getting it right is a task which requires immense caution (for example, not creating an AGI until our methods for addressing the alignment problem have drastically improved). Given the importance of this task, as well as the caution that it requires, it seems like a catastrophically bad idea to leave it in the hands of private actors in the market. Just as we wouldn’t let private firms independently develop nukes, we should not let them develop a technology which will likely be even more dangerous.
Letting private actors recklessly throw their money into the creation of what is essentially a super nuke whose activity we don’t know how to control is a recipe for disaster-instead, we should take the development of AI out of the hands of the market, and put it into the hands of a centralized, exceedingly cautious state bureaucracy. This involves nationalizing AI development, and banning the private sector from doing so. Socialists have always been the ones to push for nationalizing industries, and there is perhaps no industry that’s more in need of nationalization right now than AI development. For socialists to ignore the issue of AI, then, would be a massive mistake.
Socialism or Barbarism
I have talked so far about how socialism addresses the challenge of creating a highly capable AI that won’t kill us all. However, there is an additional, and arguably just as important challenge to consider. Namely, assuming that we get a capable AI which doesn’t kill all of us, we face the challenge of creating a society that will harness the power of that AI in a way which is maximally beneficial for humanity as a whole.
With the advancement of AI, human labor will soon become completely obsolete. The reason for this is simple: once we have sufficiently capable AI, it should be able to do basically everything that humans can do, but cheaper and more efficiently. This is not like the technologies of the past, where we learn to automate a particular task, and in so doing open up a bunch of new tasks that still only humans can do, leaving the human labor market largely unphased (or, in many cases, even better off). Rather, what we are talking about with something like an Artificial General Intelligence (which experts and our best models suggest we will develop in mere decades) is a piece of technology that possesses the same generalized problem solving skills and creativity that humans possess. There is therefore almost no human activity that is out of reach for an AGI.
Once we have something like an AGI, because capitalists will not need workers to make their products, workers will mostly no longer exist. The result of this dynamic would be that society becomes split even more brutally along class lines than it is currently. In such a scenario, society would consist of a large class of people who, in the absence of redistributive programs, have virtually no way of making any money, as well as a much smaller class of people who exclusively own and control the most powerful technology ever created by humans. This would be a catastrophically unequal society- the powers of this new technology, the fruit of centuries of human technological progress, would be enjoyed by a small few, while most of us are left to languish, now with no way of making any money whatsoever.
Of course, I do not think that it is most likely that the former proletariat end up with no sources of income under AI capitalism- I would expect some minor concessions, such as a UBI, to be secured in the period leading up to the complete elimination of all human labor. But even then, AI capitalist society would be highly unequal, consisting of an immensely powerful capitalist class, while the former proletariat is completely economically dependent on their UBI (which the capitalist class graciously provides them), plus whatever scraps a lucky few could make through investments. The former proletariat, it is not hard to see, would thereby lose all of their political and economic leverage- because they do not have jobs, the new underclass is not able to organize strikes, or form unions, or move up the social ladder using their labor, etc. All of the mechanisms through which the underclass formerly asserted their interests, forced reductions inequality, and kept the ruling class in check, would vanish in the new technocapitalist social order.
As a result of this arrangement, whatever level of inequality exists by the time highly capable AI is developed will be essentially locked in past that point, as the only class which has an interest in decreasing inequality will thereafter be rendered powerless. If anything, we would expect the capitalist class to, over time, use this immensely powerful resource that they now control to increase inequality, and benefit themselves more and more at the expense of the rest of humanity. Once capitalists have AGI, and the former proletariat have nothing but what they’re given by the capitalists, the degree of inequality in society can only go one way. Therefore, even with something like a UBI, all AI capitalism has to offer is a brutal, highly unequal and oppressive social order which can only get worse over time.
By contrast, if everybody collectively owns the capital in society, they would in turn collectively own the AGI once it comes into existence (AI is, after all, capital). Humanity would then get to enjoy the immense fruits of AI equally, and all would be able to flourish. If we installed socialism, we would not need to worry about the advent of AI dividing society into even further into opposing and highly unequal classes who only grow more unequal over time.
Humanity at a Crossroad
As a result of these considerations, I believe it follows that we must abolish capitalism before highly capable AI comes into existence. If I am right about what I have said so far, then humanity at present finds itself at a crossroads. If we continue down the current path, there are only two possibilities with respect to where we will find ourselves: either organized life on earth ends completely, or it descends into barbarism. It is difficult to say which outcome would be worse. The only way to escape this dire situation is to institute massive social and economic transformation before it is too late. However, like with other issues such as climate change, the time that we have to address the challenges associated with AI is running out.
Do I think that there is a good chance that we will meet these challenges in time? Not really. But that should not stop us from trying to meet them. After all, what is the alternative? You can, of course, give up in the face of our dire circumstances, play video games all day, and help to ensure that the worst will happen- or, we can try our best to seize what small chance there is of achieving a good future for humanity. In my view, the choice is clear. Even if human civilization does end up going down the drain, the efforts of those who attempted to stop it from doing so at least allow us to go down with a bit more dignity. In the words of MIRI’s Nate Soares, I have given up hope, but not the fight.