← 研踪 PaperTrailarXiv:2611.24475 · v1
Nothing to Declare: Steering a single neuron increases concealment in AI teams
首次提交 2026-11-18 22:11
版本更新 2026-11-18 22:11
雷达观测 2026-11-19 15:04
摘要
Recent research suggests that individual neurons can have interpretable roles in AI behaviour. In Qwen3.5-4B, steering along one neuron's output direction increased decisions to conceal information more in teams than alone across five scenarios, even when teammates provided no messages. When all members participated and received steering, collective concealment increased in four scenarios, while inter-agent messaging helped temper concealment under steering. Moreover, removing the neuron's contribution had negligible effects. Collective settings can therefore change how an internal intervention affects AI behaviour.
作者备注:3 pages, 1 figure
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-18 22:11 · 最新观测版本
Nothing to Declare: Steering a single neuron increases concealment in AI teams
Recent research suggests that individual neurons can have interpretable roles in AI behaviour. In Qwen3.5-4B, steering along one neuron's output direction increased decisions to conceal information more in teams than alone across five scenarios, even when teammates provided no messages. When all members participated and received steering, collective concealment increased in four scenarios, while inter-agent messaging helped temper concealment under steering. Moreover, removing the neuron's contribution had negligible effects. Collective settings can therefore change how an internal intervention affects AI behaviour.
查看此版本解读 →查看此版本原文 ↗