Productivity, Not Panic, Should Guide Our AI Future

A closer look at the people, practices and public choices shaping the next phase of artificial intelligence.

The debate about AI productivity needs a clearer unit of measurement. In the fictional public-service pilot described here, the team tracks time to a correct answer rather than time to a first draft. That exposes errors that otherwise disappear behind attractive statistics about output.

The results also depend on who is consulted. Front-line workers recognise awkward cases that managers may overlook, while service users identify explanations that remain confusing. Their feedback helps decide which tasks deserve automation and which are better improved through simpler forms, clearer rules or additional training.

Illustrated inaugural edition · Fictional editorial copy