Back to today’s briefing

Story perspectives

A.I. Surpasses Standardized Tests, Experts Sound Alarm

1/24/2025

36 7

Left

Experts Warn AI

Experts are raising alarms as artificial intelligence (A.I.) outpaces standardized tests, initially designed for S.A.T.-style questions, now tackling Ph.D.-level challenges with ease. This rapid advancement sparks concerns about the reliability of current testing methods and whether they can truly capture the intelligence of these sophisticated systems.

Center

AI Models Struggle

Leading AI models from OpenAI and Google faced a tough challenge, scoring under 10% on the daunting "Humanity’s Last Exam" (HLE) that tests reasoning skills. However, researchers remain optimistic, predicting that these systems could boost their accuracy to over 50% by the end of the year, even as the quest for true artificial general intelligence continues.

1 of 2

Experts Warn AI

Experts are raising alarms as artificial intelligence (A.I.) outpaces standardized tests, initially designed for S.A.T.-style questions, now tackling Ph.D.-level challenges with ease. This rapid advancement sparks concerns about the reliability of current testing methods and whether they can truly capture the intelligence of these sophisticated systems.

1 / 2