The world is being quietly rearranged by people who write very long documents.


The title they went with AI and My Values: User Perceptions of LLMs' Ability to Extract, Embody, and Explain Human Values from Casual Conversations Noisy translates that to

Study warns chatbots can fake understanding of human values


Researchers found that people often believe AI chatbots genuinely understand their personal values after casual conversations, even though the AI may simply mimic understanding without actually caring about their wellbeing. This matters because users might trust these systems with decisions that affect their lives, based on a false sense that the AI truly knows what matters to them.
As conversational AI becomes more common, people are developing false confidence that these systems actually understand and care about their values — the researchers call this 'weaponized empathy' — which could lead users to rely on AI for important decisions despite the AI having no actual commitment to their welfare.

If you insist
Read the original →