Practical Guide — SEAM + Antenarrative
AI data centers consume ~1.5% of global electricity, projected 3-4% by 2028. Communities are changed under NDA — water tables depleted, grids strained, wildlife corridors fragmented — via shell LLCs Project Bigfoot (Meta) and Project Loon (Google). This page gives you procurement-ready method from The Consent Machine.
Energy → Chips (NVIDIA) → Cloud/Data Center → Foundation Model → Applications. Each layer invisible to layer above. Change readiness tool sees application, not energy or NDA.
Require vendor to provide: (1) hidden cost baseline qQFi, (2) joint diagnosis with workers doing work, (3) internal capacity transfer plan (not dependency), (4) antenarrative audit at midpoint, (5) qualimetric tracking of rework, breach exposure, pilot abandonment, fee overrun, potential creation. No AI change goes live without someone whose job is to ask people doing work what tool gets wrong, and keep asking after go-live.
| Item | Your Estimate | Hidden Cost Signal |
|---|---|---|
| Visible consulting fees | $ ______ | ~37% of total real cost |
| AI-induced rework hours/week | ______ hrs | x salary conversion = hidden |
| Shadow AI use personal accounts | 68% orgs (TELUS 2025) | Breach exposure $4.74M avg |
| Abandoned pilots last year | ______ | Sunk cost + morale |
Full method in The Consent Machine Chapters 4-5 and Boje (in review) Ghost Vortex Capture.