AI matters but not in the way you think

Pathfinding is often associated with AI, because the A* algorithm and many other pathfinding algorithms were developed by AI researchers.

https://theory.stanford.edu/~amitp/GameProgramming/AITechniques.html

Is this AI?

Dijkstra's algorithm, one of the most well-known, foundational pathfinding algorithms in computation, was conceived in 1956. In practice, "AI" algorithms have been powering computer programs, commonly as part of video games, for decades.

Sometimes in glitchy ways too. Like Oblivion: https://www.youtube.com/watch?v=jrlzlaHEaB0

And people can find this charming and entertaining.

What about generative AI? Generative AI, or gen AI for short, is often what people mean nowadays when they talk about the rise of AI. Gen AI is a classification of AI that is, generally speaking, based on feeding a model a prompt and getting back its best guess. This can go badly wrong at times, sometimes to a comical degree. So why aren't people laughing more when it goes wrong? Why is the mood so grim?

The western internet met the western capitalist AI hype machine sometime around the attempt at pushing NFTs. NFTs were a truly disgusting capitalist development, nothing more than a means of money-changing hands, soulless and cynical profit-making that helps no one. People pushed back hard, mocked relentlessly, and NFTs seemed to recede. I won't claim this is an exact historical timeline of the events, but I'm trying to get at a rough picture here. Similar hyping going on, similar response from the internet, but then... two different outcomes.

Gen AI did not recede and has had a much more bumpy path of gains and losses along the way. The path of it, in other words, does not reflect internet reaction alone, but rather reflects material struggles. The competitive nature of capitalism put gen AI in a race toward dominance. Earlier on, it was more chaotic. People hardly knew what to try. Output quality was easily mocked as incredibly bad, rarely decent. Gradual breakthroughs changed that (quantitative to qualitative). Turning points like the Chinchilla paper led researchers toward longer training times on smaller models, to get a lot more out of less size.

The big corps took this and ran with it. Using their bottomless budgets, they could throw obscene amounts of GPU power at model training and inference. They don't have to be efficient, so they aren't. They don't have to think long-term about the environmental cost or the disruption of communities when throwing up data centers, so they don't. Make no mistake, this is capitalism in action.

The form of AI, as pointed at earlier, is not singular. It includes such things as pathfinding. Even the form of gen AI is not singular. Was it reckless when gen AI was a few researchers working on diffusion technology that eventually became StableDiffusion? Was that small-scale research costly to the environment and taking people's jobs? I don't think so. So is gen AI bad or is capitalism bad? Is gen AI bad or is imperialism bad? Is gen AI bad or is colonialism bad?

Is low quality output from an AI algorithm inherently annoying or is it annoying by association? Are people wrong to get joy from laughing about Oblivion instead of getting angry and calling it "AI slop"? Or is the difference because Oblivion isn't taking jobs and it isn't requiring massive data centers to power it?

Keep in mind, an unstable jobs economy is a trait of capitalism. Cavalier behavior with ecology is another trait of capitalism. These things were happening before gen AI came into the picture. Have we forgotten how fossil fuel companies paid money to bury and slander research on climate change? Did history begin yesterday?

This is not to say these things don't matter, of course. Of course they do matter and it's a significant part of why communism is the answer, why AES states are more going green while the imperialist west drags its feet for the nth time, why you can read about China ruling in favor of a worker over an employer trying to excuse replacing them with AI.

This is a long-term marathon fight. We can't afford to languish in the short-term capitalist point of view. It is our task to make distinctions, even when it's annoying and plodding to do and gets us called nerds. It is our task to work out how to navigate the overwhelming complexity being thrown at us and use the tools available to dismantle the exploiting system.

AI is a powerful tool. It was when it was considered pathfinding and it is when it's considered text generation. AI is a subset of automation and automation as a field isn't going anywhere. People are scared and stressed and struggling, I don't expect people to always be calm.

But try to remember to zoom out. Remember what the struggle is about and who it's for. And remember that communism will win. But we can't be afraid of tools if we're going to do it. Instead, we must return to who controls them and the form that they take.

Author's Note: This is my one allotted clickbait title for the year, you can hold me to that. :P