// AI
Scientists in an Anthropic essay say AI agents should hand back many plausible models, not one best fit
Researchers from Tübingen argue an agent that returns a single model hides equally good alternatives. Their method weighs a whole population of candidate simulators.