Why this matters

People adopt AI for the wrong tasks: the interesting ones. The wins are in the dull, repeated, low-judgement work.

The 20-minute experiment

1. 1Write down everything you did this week that took over twenty minutes. Be honest, including the admin
2. 2Mark each one: does it repeat? Does it need your judgement? Would you mind if it were done adequately rather than beautifully?
3. 3Circle the three that repeat, need little judgement and only need to be adequate
4. 4Pick the one you dislike most and do it with AI next time it comes round

Here's a week of my tasks. Sort them by how well AI could handle them, and tell me which single one to start with and why.

• A full week of tasks written down
• Each marked for repetition and judgement
• Three candidates circled
• One chosen, with a date next week when you'll try it