95% of generative AI pilots fail (MIT)
Misquoted Confidence: medium
Primary source found, but the way it is quoted changes the subject, metric, population or precision.
- Commonly cited as
- "MIT study: 95% of generative AI pilots are failing"
- True origin
- MIT NANDA (Challapally, Pease, Raskar, Chari) — The GenAI Divide: State of AI in Business 2025 (July 2025, preliminary findings)
- Type
- Stat check: A widely repeated statistic traced back to its origin.
- Year
- 2025
- Data as of
- Report dated July 2025; research period January–June 2025
- Last checked
- What the source says
Despite $30–40 billion in enterprise investment into GenAI, this report uncovers a surprising result in that 95% of organizations are getting zero return.
- Method & sample
- Review of 300+ publicly disclosed AI initiatives, interviews with representatives of 52 organizations and survey responses from 153 senior leaders at four industry conferences; labelled preliminary.
- Drift between quote and source
- Wording: '95% of organizations getting zero return' became '95% of pilots fail'
- Definition: success means a 'marked and sustained' productivity or P&L impact as remarked by users or executives
- Sample: small, conference-based, not representative
- Interest
- n/a (academic research project).
Safe rephrasing
A preliminary 2025 MIT NANDA report, based on 52 interviews, 153 survey responses and 300 public cases, said 95% of organizations were getting no measurable return from GenAI; it did not measure a pilot failure rate.
Sources we opened
Machine-readable: /api/claims/mit-95-percent-genai-pilots-fail.json · Last checked 2026-10-16