September 23, 2025Open Access

On the Psychology of a Large Language Model

View Full Paper

Ask AI

Bookmark

Mischaracterizing LLMs with human terms leads to flawed threat models and misguided safety evaluations.
The argument deconstructs LLMs as statistical systems, highlighting probabilistic pattern-matching over genuine understanding.
Philosophical perspectives define moral agency, emphasizing the absence of affective sentiment and autonomy in LLMs.
Connection of LLMs to real-world functions raises significant risks, necessitating clearer communication about their limitations.

Authors

Discussion

No takes yet. Share an insight, caveat, or question.

Analysis reveals anthropomorphism in LLMs mischaracterizes technology, indicating flawed safety evaluations.

Diogo Luiz Alves de Araújo (2025) studied this question.

View Full Paper

Ask AI

Bookmark