“The "Humanity's Last Exam" benchmark was created through a global effort where postdocs and professors contributed difficult questions designed to challenge existing AI systems.”