Choose your field.Turn the dial or pick from the list. Ten minds will take their seats.
This browser cannot draw the room in 3D, so the council is laid out as a list below.
The council, in list form
Pick a field and the ten minds appear here as well, with the reason for each rank and the tome to open.
How the ten are chosen
The twenty fields are the largest disciplines of machine learning, AI and humanoid robotics, measured by the number of distinct researchers publishing in them. The count comes from 36,974 accepted papers at sixteen flagship venues of 2024–2026 (NeurIPS, ICML, AISTATS, CVPR, ICCV, ACL, NAACL, EMNLP, COLM, AAAI, IJCAI, ICRA, IROS, CoRL and RSS), using Paper Copilot's open paper lists. The fourteen largest ML and AI fields and the six largest humanoid-robotics fields make the twenty; the dial sizes each one by its researchers.
The minds come from the first ten tomes of the Encyclopedia of Lost Minds: Echoes on AI, two hundred chapters from Gilgamesh to Malik ibn Anas. For every field, the two hundred chapters were scored against the field's vocabulary, the candidates were read, and ten were ranked by how directly each mind's own idea, and the architecture the chapter builds from it, bears on a problem the field is working on today. No mind sits on more than three councils, so the two hundred seats go to 142 different minds.
The reason given for each seat paraphrases the chapter and the historical record. The rank is a judgement, not a measurement: it says how much a researcher in that field is likely to take from the mind, in order.
| Field | Researchers | Papers |
|---|
Researchers are distinct author names on accepted papers matching the field; a paper can count toward more than one field. Fields marked with a dot are humanoid-robotics disciplines. Sources: Paper Copilot paper lists (github.com/papercopilot/paperlists); Stanford AI Index 2026 for the cross-check.