Artificial Intelligence and the Impossibility of Purity
Earlier this week I read a text arguing that artificial intelligence is fascist — not in the lazy internet sense where the word is thrown around as a generic insult, but in a structural sense: concentration of power, surveillance, exploitation of intellectual labor at industrial scale, alignment between large corporations and military interests, destruction of livelihoods, and the reproduction of the ugliest prejudices already embedded in society.
And honestly, many of those criticisms are legitimate. Palantir exists; AI companies scrape billions of pages without permission; older models associated “criminal” with Black people and “good person” with white people with disturbing naturalness. Part of the technological infrastructure of the planet is being built by companies whose only real commitment is infinite growth and market capture.
I agree with much of that criticism, but something about the entire debate has been bothering me for days, and it took me a while to understand why.
My impression is that the discussion around AI has been hijacked by moral caricatures: on one side, evangelists promising post-human abundance, infinite productivity, and a magical future where machines solve every problem civilization has ever created; on the other, people treating any use of AI as automatic complicity with exploitation, fascism, environmental destruction, and cultural collapse.
Meanwhile, almost nobody seems interested in the actual technology.
After all, “AI” has become a word that means everything at once:
- ChatGPT;
- Deepfakes;
- Spam;
- Molecular discovery;
- Layoffs;
- Datacenters;
- Ghibli-style anime images;
- Surveillance;
- Machine translation;
- SEO sludge;
- Protein modeling;
- Automated support systems;
- Medical research;
- Political propaganda;
- Accessibility tools;
- Worker replacement;
- Scientific assistance;
- Bots;
- Warfare;
- Art;
- Scraping.
All of this gets thrown into the same bucket as if it were a single coherent and conscious entity, but it is not.
There is the model itself, the data used to train it, the companies building it, the ways it is deployed, and the public narrative surrounding it. Compressing all of that into one undifferentiated mass produces far more hysteria than understanding.
When a model associates “criminal” with a Black person, it did not spontaneously invent racism. Instead, it learned patterns already present in the human-generated data it consumed. That does not absolve the technology, but neither does it transform applied mathematics into a metaphysical demon. AI amplifies social structures that already exist because it was built on top of them. The mirror does not create the face reflected in it, even if it can distort, magnify, and perpetuate it.
And perhaps that is exactly what makes this so disturbing — AI has also exposed a contradiction that has existed for a very long time, but one we were still able to pretend not to see: almost everyone materially depends on systems they consider morally questionable.
People who refuse to use AI often remind me of another figure I have encountered throughout life: the person who transforms certain refusals into a public moral identity because they possess enough economic safety to do so. I am not saying that refusal is illegitimate — in many cases it is coherent and admirable; the problem is pretending that those decisions exist in a material vacuum.
Not everyone can afford to say no.
Some people avoid openly criticizing political groups because they depend on public-sector jobs; others do not expose abuses because they understand what political persecution and harassment look like in the real world. A former client of mine was threatened by militia-connected extremists because of criticism he made during an investigation involving Bolsonaro allies. I myself can afford to say certain things because I have relative stability, professional autonomy, and contracts that protect me economically. If everything collapses in the local market, I still get paid in dollars by Canadian clients.
That is a privilege too, and this is where the discomfort becomes more serious, because I eventually realized that I am not outside the thing I criticize.
I have criticized people who perform moral purity while remaining dependent on systems that are no less destructive. But I also choose my battles and negotiate my own limits; I also make practical concessions in the name of economic stability.
I work with Israeli universities without turning it into a daily moral dilemma because, honestly, I do not want to risk losing an important contract. I use AI extensively, invest in hardware to run local models, and have dramatically increased my productivity through these tools. I have also let people go because AI could perform certain tasks better, faster, and at a fraction of the cost.
Does that make me a hypocrite? Maybe partially, maybe inevitably, or maybe simply human inside a system that transforms survival into permanent moral negotiation.
And I think this is precisely where the debate around AI loses depth when it collapses into the simplistic fight between “neutral tool” and “fascist technology.”
AI did not emerge on a healthy or just planet. It emerged within platform capitalism, the concentration of big tech power, the precarization of intellectual labor, the logic of algorithmic surveillance, and corporate obsession with infinite growth. Expecting it to arrive pure was naïve.
But it also feels simplistic to pretend that AI can be reduced to spam generation, layoffs, and fake Pixar-style portraits. There is medical research being accelerated through machine learning. There is molecular modeling, accessibility, translation, and the automation of miserable, repetitive tasks. There are ordinary people now able to produce, study, and learn at scales that were previously impossible.
The problem is that the internet transformed all of this into a religious war — perhaps because AI touched something far more intimate than previous technologies did. Social media spent years destroying people’s attention spans, but still felt like “platforms”; algorithms were shaping human behavior long before chatbots; industrial automation was eliminating jobs long before LLMs existed. But generative AI entered directly into language, writing, reasoning, creativity, and intellectual labor. It touched ego, identity, and people’s sense of usefulness. Maybe that is why so many discussions about AI take on an almost spiritual tone.
In the end, I do not think the most interesting question is whether AI is good or fascist.
The more honest question may be this: what do we do once we realize that material survival, economic comfort, personal ethics, and participation in morally questionable systems were never truly separate things?
Because perhaps the most disturbing aspect of AI is not technical at all, but the fact that it removed the disguise.