most AI training treats people like they need to be sold on AI. usage spikes when you mandate it and drops when you leave. adoption holds when tools are built into day-to-day work, and when people learn from peers they already trust.

the people holding an organisation together rarely show up that way on an org chart. network research usually finds the same thing: a small group of people quietly connecting separate silos and carrying the delivery risk for everyone else.

when teams merge or scale fast, friction gets treated as an attitude problem. underneath, it is usually two distinct histories of how good work happens being consolidated into one power structure — and aggregated data hiding who is actually progressing.

when teams stop getting along, the instinct is to coach the individuals. most of the time, the people are fine and the design is wrong: unclear ownership, competing incentives, or fatigue from repeated restructuring.

hiring bottlenecks and pay inequity aren't rites of passage. they happen when an organisation scales without redesigning how people join, progress, and get paid for capability rather than tenure.

inclusion doesn't start with outreach after the fact. criticism, curation, and community work start earlier: questioning the ethical assumptions baked into a system, and building spaces across art, science, and technology that don't require people to shrink to fit.