The Good, The Bad And The Ugly Of Open Source AI
summary
Argues that open source AI rebounds in 2025 with Chinese labs (DeepSeek, Alibaba Qwen, Moonshot AI, Z.AI) and releases like Grok 2 and GPT-OSS, critiques licenses and promotes the Linux Foundation’s Open MDW, warns that benchmarks are being cherry-picked and training-data contamination causes model collapse, and shows tooling shifts to intent-based coding via Windsurf and Cursor and GitHub Copilot plus autonomous agents—concrete examples, legal gaps, and systemic risks rather than high-level cheerleading.
AI-written summary · about
chapters (11)
- 0:00Welcome to open source AI
- 0:26Defining open source AI
- 2:38Open research in 2025
- 4:29Licensing challenges
- 6:12Intent-driven programming risks
- 6:55AI benchmark problems
- 8:19AI co-pilots to agents
- 9:23Data crisis and model collapse
- 10:49Synthetic media implications
- 11:33AI's impact on jobs
- 12:14Conclusion
related talks
16mKeynote: Open Source for AI and AI for Open Source
22mThe Imperative for Open Source AI - The Path Forward to Take Down the AI Oligarchs
39mA Hopeful View of Truly Open Source AI
26mOpen Source is Dead
27mKeynote: Open Source in the Age of AI
48mOpen Source: What's Next
42mThe Open Source AI Dilemma: Crafting a Clear Definition for Open Source AI
17mThe Future of AI, from an Open Source Perspective