AI HYPE IS BUILT ON FLAWED TEST SCORES

Summary from the TLDR AI newsletter

A growing number of researchers want to overhaul the way LLMs are assessed, calling for more rigorous and exhaustive evaluation, as they believe the practice of scoring machines on human tests is misguided.