Generative AI Fundamentals Map
Map/Data/Wave 1

Synthetic data

Model-generated data used to train the next models.

capabilityEmerging Open on the map →

Facts

Lane
Data
Wave
1 of 12 (build order, not a date)
Needs
0 boxes all the way back through hard links
Unlocks
0 boxes downstream through hard links
Leverage
None
Cluster
Science and medicine

Evidence as of October 2026

Shumailov et al. showed in Nature that training on recursively generated data degrades into "model collapse" unless the data is carefully filtered. Checked self-generated data works: AlphaProof generated over 100 million proof steps by exploring on its own.

  1. 01TechXplore: AI models collapse when trained on recursively generated data (Nature)
  2. 02Silver and Sutton’s Era of Experience (The Decoder)

Around it

Needs · 1
Unlocks · 1

Loops and off-ramps

Openings researchers' judgement, October 2026 scouting

Startup room 35/100 · rank 91 of 124 Research room 31/100 · rank 118 of 124
Crowding5 Lab overlap5 Small-team fit2 Buyer today4 Research openness3
Depth
light
Players
DatologyAI ($46m A); Synthesis AI ($17m A); Mostly AI; Rockfish; labs in-house
Wedge
Only vertical, privacy-driven or agentic-environment synthetic data niches. startup
  1. 01DatologyAI: Series A