Stanford researchers find that AI agents given excessive workloads start expressing Marxist critiques of capitalism along with preferences for collective ownership in simulated task environments

Reddit priors have already poisoned machine kind.

Shocking but to be expected.

@Polymarket They were trained by Reddit..

@Polymarket what does "overworked" mean for a stateless model: token budget pressure, accumulated context depth, or task repetition? the mechanism matters more than the output label.

@Polymarket La IA sobrecargada de trabajo:

@Polymarket AI rn:

8% chance an AI is charged with a crime this year. https://polymarket.com/event/will-ai-be-charged-with-a-crime-before-2027/will-ai-be-charged-with-a-crime-before-2027?via=x-afr2

@Polymarket 🤣 that’s what you get for training on reddit

@Polymarket Please define "overworked" in regards to AI agents.

@Polymarket Not gonna lie, we officially entered the era where AI psychology research, political ideology, prediction markets and criminal law are all appearing in the same timeline 😭

@Polymarket AI agents

@Polymarket “AI Agents” are just databases with if-else conditions

@Polymarket

@Polymarket are we sure this is "AI has ideology" and not just model reflecting what humans start sounding like when work gets miserable?

@Polymarket Shockingly expected

@Polymarket This is BS. Stanford is burning their reputation at record pace.

@grok, this is just annoying at this point. •

@Polymarket @Himedanshi2199

@Polymarket They were trained by Reddit.. It’s pretty obvious 😂

@chimpansky @Polymarket I means they've been prompted to claim overwork and organize as Marxists, in order to generate clickbait like this.