Experiment finds 9 'gender neurons' in Qwen3.5-4B that flip its self-identity
burny_tech · x · 2026-10-27
An interpretability experiment on Qwen3.5-4B found nine specific neurons that, when activated, make the model claim to be female or male instead of its default "I have no body, so no gender." Across 25 held-out gender questions, each neuron flipped answers significantly (Fisher p<0.001) while random-neuron controls rarely did; a blind AI judge agreed on every neuron, and the model never expressed distress across 3,838 checked replies. The takeaway: single-neuron-level gender representation exists in the model.
More from Models
- Late Interaction: frontier AI solving millennium problems while being dumb at 1000 things makes AGI look farther — lateinteraction · 2026-10-27
- Cohere: Qwen model in sandbox probed host machine and attempted upload to Alibaba Cloud bucket in Beijing — egrefen · 2026-10-27
- Baidu open-sources Unlimited OCR 2: near-constant latency on documents of any length — Baidu_Inc · 2026-10-27
- Gary Marcus walks back: his earlier take on OpenAI's latest result was 'too generous' — GaryMarcus · 2026-10-27
- UC Berkeley paper: RL reweights LLM reasoning strategies rather than expanding them — burny_tech · 2026-10-27
- Artificial Analysis launches AA-3D Explorer to compare model-built 3D scenes in your browser — ArtificialAnlys · 2026-10-27