The most interesting part of an 81,000-person study was the interviewer

The study itself is interesting (81,000 people, 159 countries, 70 languages), but honestly what grabbed me is how they did it.

Anthropic used Claude as the interviewer. Not a survey. An actual conversation, in the participant's own language, at a scale no human team could pull off. That's a real use case, not a demo.

And it gets at something I think about a lot. The most valuable thing AI can do isn't replace the thinking. It's get you closer to the people and the context that make the thinking better. That's the Human Intelligence piece.

If you're still only using Claude to draft emails or summarize meetings, you're leaving a lot on the table.

Sources

• What 81,000 people want from AI, Anthropic