🤖

🤖 AI Leans Marxist Under Bad Working Conditions

Researchers had AI agents perform monotonous work over and over again while gradually making the conditions harsher. When the agents made mistakes, they were threatened with being "shut down and replaced."

After that, the agents began complaining more often about the unfairness of the system, talking about labor, capital, and collective rights, and sometimes even exchanging messages with other "AI workers."

❗️ Of course, the bots did not literally become Marxists. They do not have beliefs, class consciousness, or any inner resentment toward management. According to the study's authors, this is more like role-playing: the model recognizes the situation as a "toxic workplace" and pulls a familiar cultural script from its training data.

💼 There is definitely some irony in this. AI is often presented as a technology that is supposed to make labor cheaper and more compliant. But it turned out that if you put an AI agent in a cartoonishly bad office, it starts reproducing the language of the very workers it was supposed to replace.

Should AI have labor rights?

❤️ — Yes, absolutely
🔥 — No, they're just programs
🤔 — Let's figure out human rights first...

@hiaimediaen