Editor’s pick
What Should an AI Be Allowed to Remember?
A product-builder’s analysis of AI memory, inferred identity, deletion, and a practical contract for governed retention.
Archive
Essays for living thoughtfully with AI.
Long-form writing on the parts of artificial intelligence that are easiest to feel and hardest to measure.
Editor’s pick
Rebuild Self-Trust With a Calibration Record
A seven-day protocol for replacing constant AI reassurance with predictions, outcomes, and selective verification.
AI Enters the Room Before Anyone Opens It
A human-first meeting protocol for protecting independent input, dissent, and named responsibility before AI compresses the room.
Editor’s pick
The Apprenticeship Problem After Generated Answers
A transfer-based method for using AI without removing the perceptions and decisions an apprentice needs to develop.
When Personalization Becomes a Theory of You
A practical argument for making AI personalization contestable through provenance, scope, expiry, correction, and influence audits.
Editor’s pick
Keep One Draft From Before the Suggestions
A three-artifact archive for preserving provenance, recording AI assistance, and making editorial responsibility inspectable.
The Second Opinion That Never Leaves
A decision memo and consultation budget for using AI to challenge choices without reopening them indefinitely.
What Are You Hiding When You Hide AI Use?
A context-sensitive disclosure test for separating ordinary assistance from misleading attribution, hidden dependency, and avoided learning.
Editor’s pick
The Work the Checklist Cannot See
A method for preserving tacit knowledge through exception records, supervised cases, and feedback on what practitioners notice.