My thoughts on AI as a socialist software developer

This blog post is a deviation from the kind of content I have historically written here, mainly because of how transformative and divisive artificial intelligence has become for this (and many other) industries.

What's happening?

Different things are happening according to different groups of people, and people construct their understanding of reality largely based on their own experiences. If we want to align on what to do, it's first important to bridge the gaps between the groups' understandings. We'll differentiate said groups by letters.

For group A, not much is happening. This group hears people talk about the end of the world, about some grand utopia, and then they go to these chatbots and see (arguably) better search engines that sometimes get important questions wrong. They conclude that this is just fear mongering by media to drive engagement.

For group B, this is a classic example of corporatism failing in America. Money is flowing inside this bubble, pumping valuations of these companies far beyond the value that they provide to people, based on the promise of a future that replaces human laborers and in the meantime vastly worsening local climate/ecological issues. Things like noise, pollution, water usage, and heat production are all massive concerns.

According to this group, the prospect of replacing human workers isn't even a good end goal. It will lead to massive suffering as workers become unemployed if the technology continues to improve, and the things we derive purpose from (coding, art, research, etc.) will cease to be human activities. If it doesn't improve, then the bubble popping will create massive harm.

For group C, this is the most incredible and important invention that has ever existed. Intelligence is nearly the sole mechanism that allowed us to shape the world far beyond what other animals have been able. If we can just let these companies compete in the free market to create, align, and harness an even more sophisticated intelligence than our own, then we could solve problems that humans have been trying to solve for decades or centuries. We could cure many more cancers, diseases, even ageing. We could construct full dive VR, creating digital utopias that allow us to experience anything we want, beyond even what our senses allow.

Who is right?

None of them are completely right, and they're all fundamentally flawed in the same kind of way. Before we get into that, it's important to note they each pull from some part of reality, so these views aren't baseless:

• group A correctly sees that chatgpt can operate as a search engine, that the media tries to drive engagement (often through fear), and that AI can be incorrect at times

• group B correctly sees the failure of private AI development harming humans, and identifies harm that the private corporations who develop this technology have already caused.

• group C correctly sees intelligence as an avenue for creating an astronomical amount of good in the world, and actually we're already seeing AI solve problems that humans have been stuck on for decades.

The main issue is these groups all develop from a liberal frame (I mean liberal here in the academic sense, not the American sense). Essentially all people who are not socialists or fascists are liberals. There are exceptions, but these 3 groups can account for easily 99%+ of humanity.

Liberalism dominates the world, so it makes sense that the most popular frames for viewing important topics like artificial intelligence are done in this frame.

What's wrong with this frame?

The liberal frame can sometimes identify the harm that it causes, but it can never suggest a solution that comes in conflict with the most important component of capitalism: the stratification of humans into different classes (owners vs workers). A suggestion that there ought to be no more workers is advocating:

• at best, that we transformer the dichotomy into owners vs non-owners, a final stage of capitalism, so to speak, where one class controls production (and benefits with massive allocations of capital), while the other class consumes on subsistence wages provided by the state (universal basic income)

• at worst, that the "unnecessary" workers be systemically denied access to basic necessities like shelter, food, etc.

For a lot of liberals, this doesn't feel like a gamble worth taking, it'd be better to develop new-age luddism, outlaw all technology post-2020, and let people go back to making the things that give them purpose, code, art, etc. This gets to a fundamental misunderstanding liberals have. Liberals often pontificate not only that we ought to derive our worth from the work we do for capitalists, but that this actually happens at an incredibly large scale.

Of course, it doesn't. Essentially anyone who works can attest, it's miserable. You lose basic autonomy over your life (what to do with your time), and the alienation from your own labor can turn what was once a passion into something that you stop looking forward to doing ("crowding out" in psychology). For a lot of people, there wasn't even a passion to begin with, it's just soul-numbing work to pay rent.

Transformation is not only necessary, it's inevitable.

Often times at this point, you might hear the critique "okay but socialism will never happen". Humans tend to view things as more stable than they are, as a result of living in essentially the same conditions for their entire life. Humans are not critical thinking machines, so when a human lives the same life for decades on end, going to work despite hating it the entire time, when someone promises change, after hearing promises for change over and over again, they've learned to internalize that it's impossible.

Human societies are not particularly stable, and I'd argue they're getting less stable as time goes on. As technological growth accelerates through the millennia, we see societies reform to align with it. We saw hunter/gatherers stable for hundreds of thousands of years, leading into many thousands of years of smaller villages, that eventually led into many thousands of years of larger city states, and then the reformation of these city states into nations that within a couple thousand years discovered technology that necessitated radical sociopolitical change worldwide to capture these new technologies for the ruling class.

Whatever path forward we take from here, society will be radically transformed.

If conservative-liberals/luddites get what they want, then for the first time in the history of humanity, we will have stopped progress even when we knew how to progress. That has never been achieved before, and I see absolutely no reason why it would work now. It would take an unprecedented level of global cooperation when no singular nation in the world has enacted this kind of pause on progress.

If progressive-liberals get what they want, we'll have constructed a pseudo-permanent two-tiered society. As capitalism stands right now, workers actually hold all the power, but owners are far better organized due to their capture of the state. If workers collectively organized a general strike, whatever demands the workers had would have to be met. A general strike post-work would do essentially nothing. Owners would finally have no working class to combat. We wold have to survive on the good will of the owner-captured state to provide us with enough money to survive.

What do socialists want?

What we've always wanted, democratic control (social ownership of the means of production). We should be aligning new technologies with human interest, not profit. It may be more profitable to teach AI to replace coders/artists, but it may have been just as easy to allocate this level of capital into robotics and instead first replace menial labor.

I do think the development of hyper-intelligent beings (or "things" if you prefer) can bring an astronomical amount of good to the world, as I said earlier. I do think there's a good argument for focusing on this kind of technology before or alignside the development of robotics.

The missing part in liberalism is the collective ownership. The development of automation is an existential threat under capitalism, that as I said can be best answered by a two-tiered society. Within socialism, automation would instead lead to the rise of the quality of life of the collective owners (all humans in society). There would be no owner class deciding to allocate AI towards helping Israel commit a genocide, instead there'd be workers deciding to allocate AI towards creating new vaccines or treatments for cancer.

Within a socialist society, the economy would be democratized the same way the state can be democratized. We'd decide how to allocate resources collectively, rather than by owners for profit.

There is a lot of debate within socialism about how this democratization should take place (through a state? federated unions?). There's debate about how the economy should actually be constructed, but there is agreement across all socialists that there ought to be no owner/elite class. We need to exist in a classless society, free from masters, where we collectively own the means of production (including AI).

Does socialism solve all the possible issues that can arise with superhuman intelligence?

Nope, no singular idea will solve all the problems that intelligence can create. My main point here is that the liberal frame is limiting us from astronomically more powerful solutions to these problems that are far more likely to work.

Profound change is inevitable, we might as well do it right.