Untitled

# Claims for Why AI Pilots Don't Become Products

## Central idea
Most AI pilots fail to reach production not because the models are bad but because organizations lack the absorptive capacity, data infrastructure, and change readiness to operate them over time.

## Claims I plan to make

1. Most AI/ML projects do not make it to production. Source type: hedge.
- How I will phrase the uncertainty: "Gartner and others have repeatedly noted that most AI projects fail to make it past the pilot phase." No specific year or percentage stated as precise.

2. Pilots succeed under controlled conditions (curated data, motivated team, clear narrow scope) that do not exist in production. Source type: hedge.
- Framed as observed pattern in industry, not a citation.

3. IBM Watson Health: large investment, many announced partnerships, disappointing outcomes, unit sold around 2021-2022. Source type: hedge.
- How I will phrase the uncertainty: "by most accounts." Based on press reporting, not internal data.

4. McKinsey-type argument that AI value requires organizational change alongside the technology. Source type: hedge.
- Framed as widely noted in industry commentary, not pinned to a specific report.

5. MLOps emerged specifically because deploying models is a different discipline from building them. Source type: hedge.
- Widely documented; stated as observable fact in the field.

6. Cohen & Levinthal (1990): absorptive capacity — organizational ability to recognize the value of new external information, assimilate it, and exploit it for commercial ends; path-dependent and cumulative. Source type: local.
- File: /Users/alisafari/Downloads/PHD/UNT/2026/COMPS/study-hub/day1.html (lines 1217, 1231, 1311)

7. An organization that has not invested in prior related knowledge cannot absorb new knowledge in a domain regardless of technology spend. Source type: local (derived from Cohen & Levinthal mechanism).
- File: /Users/alisafari/Downloads/PHD/UNT/2026/COMPS/study-hub/day1.html (same lines)

8. Technology readiness is multi-dimensional: infrastructure, data pipelines, people, processes must all be ready simultaneously. Source type: hedge.
- IS theme; framed as IS literature observation, not a single citation.

## External URLs I will cite
None — all external claims are hedged industry observations, no specific URLs.

## Internal links I will use
- /blog/absorptive-capacity-why-tools-alone-fail (absorptive capacity mechanism applied to technology adoption)
- /blog/erp-implementation-failure-structural-not-technical (organizational resistance patterns in enterprise IT)

## Sources I read while preparing
- /Users/alisafari/Downloads/PHD/UNT/2026/COMPS/study-hub/day1.html (lines 1217, 1231, 1246, 1311 — absorptive capacity section)
- /Users/alisafari/Downloads/PHD/UNT/2026/COMPS/alisafari-space/blog/2026-05-14-absorptive-capacity-why-tools-alone-fail.md (voice reference and existing treatment)
- /Users/alisafari/Downloads/PHD/UNT/2026/COMPS/alisafari-space/blog/2026-05-14-erp-implementation-failure-structural-not-technical.md (voice reference)