AIs like ChatGPT fall apart in classic 'Stroop' psychological test — and that could stand in the way of achieving artificial general intelligence
SMRTR summary
A new study tested ChatGPT and Claude on the psychological "Stroop test," where AI had to name ink colors of mismatched color words. GPT-4o's accuracy collapsed from 91% to 15% as word lists grew longer, while humans stay near 95% accuracy. Researchers say fixing this attention problem is essential for achieving true artificial general intelligence.
SMRTR provides this summary for quick context. The original article belongs to TechRadar.
Read the original article