| title | MIT CISR — Four AI Maturity Stages |
|---|---|
| source-id | mit-cisr-2024 |
| wiki-page | (02_References/enterprise-ai/wiki/frameworks/mit-cisr-maturity-stages.md) |
| last-synced | 2026-06-07 |
MIT CISR's empirical four-stage AI maturity model (721-firm 2022 survey, updated 2024). Stage assignment is based on Total AI Effectiveness. Firms in Stages 3–4 financially outperform their industry; firms in Stages 1–2 underperform. The Stage 2 → Stage 3 transition is the critical inflection point.
| Stage | % of firms | Focus | Revenue vs. industry |
|---|---|---|---|
| 1. Experiment and Prepare | 28% | Literacy, acceptable-use policy, data accessibility | −12.6 pp |
| 2. Build Pilots and Capabilities | 34% | Business cases, tracked pilots, APIs, value measurement | −3.5 pp |
| 3. Develop AI Ways of Working | 31% | Shared platform, architectural reuse, test-and-learn culture | +11.3 pp |
| 4. Become AI Future Ready | 7% | AI in all decisions, AI-as-a-product to external clients, four-modality combination | +17.1 pp |
Source: MIT CISR, 2022 survey.
Stage 1: Has literacy and acceptable-use policy, but no tracked, value-attributed pilot. Stage 2: Running tracked pilots with named business cases and measured value — but value is contained to single functions; no shared platform. Stage 3: Shared AI platform with reusable services; dashboards expose AI value; test-and-learn is the default operating mode. Stage 4: AI in default decision-making across functions; monetizing AI capability externally; combining all four AI types (analytical, generative, agentic, robotic).
"Enterprises in stages 1–2 had financial performance below industry average; enterprises in stages 3–4 had financial performance well above industry average." — MIT CISR
Stage 2 → Stage 3 requires a platform investment that ROI-driven Stage 2 cultures resist (no individual pilot pays for it). This is the most common stall point.
- DBS Bank: S$150M AI impact (2022) → S$370M (2024) → S$1B+ expected (2025). 350+ use cases. CEO KPI: 1,000 experiments/year.
- Ping An Insurance: 49% of product sales by AI representatives; AI banking as a service to 30 other Chinese banks.
- For skill
general-maturity-assessment: MIT CISR's four stages are the primary sequential axis of the assessment. Plot the org on this axis first, then use Accenture's 2×2 to identify the binding constraint. - For skill
tech-stack-diagnostic: Stage 3 platform requirements (shared AI platform, architectural reuse) are what the stack diagnostic should be building toward.
MIT CISR Enterprise AI Maturity, 2022 survey + 2024 interviews. 721 firms; Total AI Effectiveness measure; four-stage model; DBS Bank and Ping An Stage 4 exemplars.