https://eoinkenny.github.io/blog/ Research notes on AI and interpretability. en https://eoinkenny.github.io/blog/show-the-judge-a-neighbour/ https://eoinkenny.github.io/blog/show-the-judge-a-neighbour/ A small probe flags a sentence. An LLM judge dismisses it. One labelled example changes the verdict. What can that reversal teach us about explanations? Mon, 26 Oct 2026 00:00:00 +0000