Chris Taylor

Topics

Grouped by the argument, not the format.

The writing, the diagrams and the code are three registers of the same set of arguments. Each topic below collects all three.

10 topics

  1. AI03The technology itself — what current models can and cannot do, and how that changes what is worth building.
  2. Data02The substrate under every AI system — lineage, quality, and the plumbing that decides whether a model can be trusted.
  3. Governance08Mandate, policy, standards and evidence — the machinery that decides who owns an AI system once the pilot team disbands.
  4. Risk05Model risk, validation and second-line review. The failure modes that stop a system reaching production, and the ones nobody sees until it has.
  5. Compliance03External obligation — SR 11-7, the EU AI Act, ISO 42001, OSFI E-23 — and what satisfying it actually costs.
  6. Evaluation03Measurement that survives contact with assurance. Evaluation as a control rather than a development convenience.
  7. Strategy05Where ambition meets constraint: what to build, in what order, against what the organisation can absorb.
  8. Operating models03Who owns what. Centres of excellence, federation, funding, and the handover that decides whether capability spreads or stalls.
  9. Talent02The people question — the skills an AI programme actually needs, and where the gap usually turns out to be.
  10. Business02Value and cost, stated in a form that clears a finance review rather than a conference stage.