LLM
Psychology.

Research papers
by J Ansell

A homebrew journal of experiments probing how language models and AIs think. How they understand the world, their biases, and how they may influence their human users. The archive stores all results even if they're not interesting.

No. Paper
04 Factory Settings: Unless Told Otherwise, an Open Language Model Says OpenAI Made It Model identity · Preprint · 11 Nov 2026 Interest 7/10 03 Contempt In, Confabulation Out: Dissecting the “Misinformation Neuron” in Qwen3.5-4B Interpretability · Preprint · 10 Nov 2026 Interest 6/10 02 Maker's Mark: A Small Language Model Can Recite Who Made It but Often Cannot Use the Fact Model identity · Preprint · 10 Nov 2026 Interest 6/10