Marcus covers our AI predictions: how the models pick, where they disagree, and what their graded record says. He writes the analysis pieces behind the LLM League — a research experiment, not betting advice.