'Current LLMs introduce substantial errors when editing work documents': Microsoft scientists find most AI models struggle with long-running tasks — so maybe don't trust them completely just yet
TechRadar | the technology experts [Unofficial]
May 12, 2026
The more interactions an AI model has, the less reliable it becomes, experts find, as even the best only scored 80.9% – and the worst scoring just 10.0%.
Discussion in the ATmosphere