Nathan Lambert and Tom Zick have launched Trillium Labs, a nonprofit dedicated to conducting high-stakes AI research openly. They argue that the current trend of frontier AI labs keeping their work secret hinders community scrutiny and risk mitigation. Trillium Labs plans to publish details of experiments, focusing on areas like recursive self-improvement and agent development, to foster a shared understanding of AI risks and capabilities. AI
IMPACT Challenges the prevailing closed-door approach to AI safety, potentially accelerating open-source development and risk mitigation strategies.
RANK_REASON Founding of a new nonprofit by prominent AI researchers to pursue a different model for AI safety research.
Read on Medium — Anthropic tag →
- Allen Institute for Artificial Intelligence
- American Truly Open Models
- Anthropic
- Charles Schwab Corporation
- Harvard University
- Hugging Face
- Nathan Lambert
- OpenAI
- Stanford University
- Tom Zick
- Trillium Labs
- University of California, Berkeley
- Xiaomi
- Claude 3.5 Opus
- Claude 3.5 Sonnet
- GPT-4
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →