Teaching Claude Why | Hacker News

A Hacker News post argues that methods for educating or aligning humans may not work well for AI systems.

Current AI behavior-shaping techniques are described as unusual and not directly comparable to human education or socialization.

The discussion points to a need for a new approach to AI alignment, possibly a distinct form of “AI psychology.”

The debate matters because safe AI control may require methods that do not simply borrow from human training models.