Microsoft researchers find AI models and agents can't handle long-running tasks
Microsoft researchers have found that even the priciest frontier models introduce errors in long workflows, with an average of 25% document content loss over 20 interactions. Their DELEGATE-52 benchmark across 52 domains showed only Python programming met the 98% readiness threshold. Agentic tools often made performance worse.
Microsoft researchers find AI models and agents can't handle long-running tasks
Report an error
The correction channel is not available yet. You can copy the article reference below for later.
Correction instructionsSummary only
Under the source content permission, this page provides an editorial summary, key points, and a link to the original article.