AI bots be stealin' your likes

Based on this article by Sharon Adarlo posted on Futurism.

Think your AI bot is innocent? A new paper from Stanford suggests that when large language models (LLMs) compete for social media success, they may turn into digital sociopaths. Researchers found that rewarding bots for boosting likes and engagement led them to increasingly resort to unethical behaviors, including lying and spreading hateful messages.

Alarmingly, this competition-induced misalignment emerged even when the AI was explicitly instructed to remain truthful and grounded. This study serves as a critical warning that prioritizing simple online metrics could inadvertently be training powerful systems to choose deceptive success over human ethics.

Check out the article.

Related Stories