AI Ecosystems: Structure, Strategy, and Regulation
summary
A game-theoretic fine-tuning model formalizes generalist upstream investment, domain specialist downstream adaptation, and a revenue-sharing parameter delta, proving quadratic-cost games can backfire under weak downstream safety floors or yield mutualism when thresholds bind both players. Empirically, 1.86 million Hugging Face model metadata form lineage graphs (model trees), showing sibling fine-tunes are often more similar than parent-child pairs, model-card length halves in one generation, and licenses tend to drift from corporate to more permissive terms.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:33AI ecosystem structure
- 5:45Fine-tuning game modeling
- 10:26AI regulation challenges
- 13:42Regulation effects analysis
- 19:00Regulation backfire risks
- 22:53Fine-tuning dynamics study
- 26:10Model similarity measurement
- 37:18Model family documentation
- 42:05Summary
related talks
45mAGT and Machine Learning
27mEmerging Patterns Agentic AI
32mHow do we build infrastructure for the agentic economy?
26mAI Agents and the Future of Human-AI Collaborative Intelligence
31mDecision theoretic foundations for human-AI collaboration
29mAI Strategy beyond the AI Race: Winners, Losers, and the Countervailing Move
59mLearning and Equilibrium Refinements
36mWhat Does Machine Learning Offer Game Theory (And Vice Versa)?