What AI Still Can’t Do for Leaders
Leaders are handing off decisions to AI. MIT researchers just mapped where that strategy breaks down.
Why it matters
As enterprises deploy generative AI into leadership workflows, academic research is identifying the specific decision domains where AI output remains unreliable and where human judgment delegation creates organizational risk. This frames AI limitations not as technical problems but as governance and leadership questions.
The key facts
10 to knowSource: MIT Sloan School of Management (Ancona & Isaacs)
Focus: Leadership decision-making gaps with generative AI
Key tension: AI capability vs. responsible human agency in org decisions
Tools referenced: ChatGPT, Claude in daily workflows
Audience: C-suite and organizational leaders adopting AI
MIT Sloan School of Management study on AI limitations for leadership
Authors: Deborah Ancona and Katherine W. Isaacs
Focus areas: ChatGPT and Claude adoption in leadership workflows
Key tension: Output gaps vs. agency delegation risks
Published: MIT Sloan Review (May 2026)
Go to the source
MIT Sloan Management Reviewsloanreview.mit.edu
Publisher excerpt: As leaders weave generative AI tools like ChatGPT and Claude into their daily workflows, where does the output fall short? Moreover, where are leaders falling short for their organizations by giving away too much agency to artificial intelligence? MIT Sloan School of Management professors Deborah…