Stroop Test Exposes Inherent LLM Flaw

Summary: A new cognitive evaluation of artificial intelligence has unmasked a fundamental, systemic flaw running through large language model (LLM) attention mechanisms. By administering the classic psychological “Stroop task” to…