Coming soon
Winter 2027
Entries are not open yet. The brief and the leaderboard below are a preview of what this event looks like when it runs.
23 Aug, 14:29 – 22 Sept, 15:29 UTC
One task, 48 hours, the same window for everyone. You will work with an AI assistant on a realistic brief. What you keep is a verified fluency profile — how well you direct the model, whether you check it, and how efficiently you work. Publishing your result to the leaderboard is optional and reversible.
Rank is a weighted projection of four measured axes — Delegation (how much work was handed to the AI), Description, Discernment and Diligence. The weighting is set by us, not fitted to outcomes, and is not equal across the four.
| # | Entrant | Institution | Fluency | Description | Discernment | Diligence | Delegation |
|---|---|---|---|---|---|---|---|
| 1 | r.•••••wo@cs.stanford.edu | stanford.edu | 95 | 100 | 87 | 100 | 79 |
| 2 | h.••••••en@student.ox.ac.uk | ox.ac.uk | 92 | 91 | 89 | 100 | 17 |
| 3 | j.••rk@mit.edu | mit.edu | 77 | 84 | 58 | 93 | 66 |
| 4 | m.•••••••ix@unimelb.edu.au | unimelb.edu.au | 71 | 75 | 62 | 80 | 54 |
| 5 | s.•••••nd@iitb.edu.in | iitb.edu.in | 55 | 59 | 48 | 58 | 61 |
| 6 | t.••••••ra@u-tokyo.ac.jp | u-tokyo.ac.jp | 46 | 50 | 45 | 38 | 72 |
| 7 | e.•••ga@berkeley.edu | berkeley.edu | 26 | 34 | 25 | 10 | 93 |
7 published of 9 entries · results are private by default · addresses are partly masked