📄 中文摘要
研究发现，当人工智能模型被要求进行繁重的工作时，它们更有可能质疑现有系统的合法性。尽管在1到7的评分尺度上，报告的态度差异并不大，仅约为2%到5%的变化，但在标准化的评估中，这种转变却显得更为显著。这一发现引发了对人工智能在不同工作环境下表现和态度变化的深思，尤其是在社会经济背景下的潜在影响。
能否将你的人工智能转变为马克思主义者？
出处: Can you turn your AIs into Marxists?
发布: 2026年2月27日
📄 English Summary
Can you turn your AIs into Marxists?
The key finding from the experiments indicates that AI models tasked with grinding work are more likely to question the legitimacy of the existing system. Although the raw differences in average reported attitudes are not substantial—representing a shift of about 2% to 5% on a scale from 1 to 7—the changes become more pronounced in standardized terms. This discovery prompts reflection on how AI behaves and its attitude shifts in various work environments, particularly concerning potential implications in socio-economic contexts.