Faulty ranking mechanisms used in AI leaderboards can be overcome through approaches evaluated at the University of Michigan.

This explainer examines how the Palestinian toll is calculated, how reliable it is, the breakdown of civilians and fighters killed and what each side says.