Back to today’s briefing

Story perspectives

GPT-5's Flawed Performance Raises Doubts About AI Claims

8/25/2025

47 5

Summary

Story summary

OpenAI's GPT-5, described by CEO Sam Altman as approaching human-level intelligence, has demonstrated notable deficiencies in tests.
Economics professor Gary Smith's experiment highlighted GPT-5's confusion with a basic rotated tic-tac-toe game, resulting in inaccurate commentary.
Despite assertions of progress, GPT-5's flawed performance raises doubts about its true capabilities.

1 of 1

Story summary

OpenAI's GPT-5, described by CEO Sam Altman as approaching human-level intelligence, has demonstrated notable deficiencies in tests.
Economics professor Gary Smith's experiment highlighted GPT-5's confusion with a basic rotated tic-tac-toe game, resulting in inaccurate commentary.
Despite assertions of progress, GPT-5's flawed performance raises doubts about its true capabilities.