We Taught AI to Answer Questions. Who Taught It to Change Its Mind?
I built a benchmark to test if AI systems can reliably update their conclusions when new evidence arrives. While all methods scored 100% when calculating from scratch, their ability to revise existing states varied drastically, with some strategies failing to update critical facts or breaking unrelated logic.