Further reading
The survey draws its figures and examples from the public record rather than from forecast. These are the sources each chapter leans on — the studies, the industry reports, and the reporting that documents the flood. They are grouped by the terrain they map, and best read as the state of a fast-moving field at the time of writing, not as fixed coordinates.
Chapter 1, The Jagged Frontier, rests on standard results from the machine-learning literature — tokenization, the scaling laws (Kaplan; Chinchilla), and circuit-complexity bounds on a single forward pass — rather than on a discrete bibliography.
02
The Content Factory
Pop culture · the synthetic floor- Futurism — AI band's streams on SpotifyThe Velvet Sundown
- Music Business Worldwide — The AI music problem on streaming platforms
- The Guardian — Spotify removes 75m spam tracks
- The Guardian — AI, bot farms and innocent indie victims
- NPR — When your favorite band's new song is an AI fake
- Neural Horizons — The Flood: AI slop as a DDoS on attentionNewsGuard data
- Vervocity — What is AI slopYouTube recommendation data
- ListenFirst — AI slop and engagement / traffic data
- Entrepreneurloop — AI slop explainedActivision fake-game ads
- iPullRank — The content collapse and AI slopAhrefs 74% figure
- Futurism — Slop farmer using AI to flood social media
03
The Machine That Must Behave
Games · the irreducibility moat- Google DeepMind — Genie 3: a new frontier for world models
- Codecademy — Genie 3 overview and limitations
- Wikipedia — Genie (AI model)
- arXiv — From Masks to Worlds: a guide to world models
- AI Buzz — AI in gaming and game development studio guidegameslop · GDC · SAG-AFTRA · Copyright Office
- The Gamer — Steam's January release flood
- SlashGear — Krafton replacing humans with AI
- Dot Esports — SAG-AFTRA video game actors strike over AI
- Wikipedia — 2024–2025 SAG-AFTRA video game strike
- GDC — 2025 State of the Game Industry
- Deconstructor of Fun — AI × game industry, bull and bear cases
04
The Specification Is the Work
Software · accidental vs. essential- METR — Measuring the impact of early-2025 AI on experienced developer productivity
- METR — The productivity study, full paperarXiv 2507.09089
- METR — Changing the productivity experiment designFeb 2026 update
- Simon Willison — Notes on the METR study
- Google Cloud — Announcing the 2025 DORA Report
- DORA — 2024 Accelerate State of DevOps Report
- IT Revolution — AI's mirror effect2025 DORA
- Faros AI — Key takeaways from the 2025 DORA reportthe AI productivity paradox
- OpsLevel — DORA 2024 throughput / stability figures
- RedMonk — DORA 2025, measuring delivery after AI
- arXiv — The Novelty Bottleneckhuman-effort scaling · the march of nines
05
The Verifier and Nature
Science · where nature is the judgeMathematics
- Nature — Olympiad-level formal mathematical reasoning with reinforcement learningAlphaProof
- UVA — Major breakthroughs in Lean 4-based auto-formalized mathematics
- Dave Shapiro — How good is AI at math, really: an anti-hype reality checkIMO · Erdős problems
- arXiv — The FrontierMath benchmark
- arXiv — The Mathematician's Assistant: integrating AI into research practice
- Terence Tao — Mathematical exploration and discovery at scale
Physics, chemistry, biology
- arXiv — AI for scientific discovery is a social problemAlphaFold · GNoME · AlphaFold3
- IFP — How X-Labs can unleash AI-driven scientific breakthroughsEvo 2 · research infrastructure
- ScienceDaily — Self-driving lab discovers materials 10× faster
- R&D World — 6 ways AI reshaped scientific software in 2025Boltz-2 · RFdiffusion · OpenFold3
- ETC Journal — AI scientific research innovations, Dec. 2025AlphaFold · GNoME
- arXiv — PhysMaster: building an autonomous AI physicistGraphCast surrogates
Read the chapters these sources support in the full survey, or see the lenses they are read through in the framework.