Data-Driven Notes.

Benedict Evans and Toni Caron-Brown argue that scoring jobs by AI exposure is “ludicrous.” They have a point. Here’s what we’re doing about it.

The site now runs in Swedish and English. French, German, Spanish, Italian, and Greek are ready in a test environment and need native-speaker validation before going live. More languages can be requested.

The site now supports French, German, Spanish, Italian, and Greek alongside English and Swedish. More languages are in the works, and we'd love to hear which ones matter most to you.

This week we mapped six new roles for the first time: Health Data Scientist, Instructional Designer, IT Strategy Manager, Police Officer, Startup, and Tennis Coach. The spread hints at where people are most curious about AI exposure.

We added a research benchmark from Eloundou et al. (2023) to every profession and results page — here is what their method is, how it compares to ours, and why we show both.

We've launched Discover, a new way to explore careers by the tasks you love rather than job titles. Start with what energizes you, and see both the fit and the AI exposure at once.

New data from 8,600+ job assessments reveals where AI tools actually create leverage today. The answer isn't about replacing workers—it's about identifying which specific tasks in your role have thin enough context requirements to automate reliably.

Six senior leadership titles entered the map this month. Early data suggests executive roles retain substantially more human advantage than specialist positions, even in data-heavy domains.

Synthetic data can seed a dataset. It can't replace the thing it's standing in for. Understanding where AI automation actually lands — not in theory, but in practice — requires the people doing the work to say what the work is really like.