Self-building benchmarks: Using AI-generated exams to understand LLM work capabilities

Maria del Rio-Chanona and Johanna Einsiedler offer a new, AI-driven approach to measuring AI’s capabilities in the workplace