In this week's MLAISU, we're covering the latest technical safety developments with GPT-4, looking at Anthropic's safety strategy, and covering the fascinating Japanese alignment conference!
- Join our Discord! https://ais.pub/discord
- Join the AI governance hackathon! https://ais.pub/aigov
- Check out the university job opportunities: https://ais.pub/opportunities
Sources
- Japanese alignment conference 2023: https://jac2023.ai/
- Recordings from JAC2023: https://vimeo.com/user196160056
- GPT-4 released: https://openai.com/product/gpt-4
- GPT-4 technical report: https://cdn.openai.com/papers/gpt-4.pdf
- Developer demo: https://youtu.be/outcGtbnMuQ
- Inverse scaling: https://www.lesswrong.com/posts/eqxqgFxymP8hXDTt5/announcing-the-inverse-scaling-prize-usd250k-prize-pool
- IQ score: https://twitter.com/DanHendrycks/status/1635706827215339520
- Why uncontrollable AI seems like a larger risk than ever: https://time.com/6258483/uncontrollable-ai-agi-risks/
- Is power-seeking AI an existential risk? https://arxiv.org/abs/2206.13353
- OpenAI evals: https://github.com/openai/evals
- Anthropic's AI safety views: https://www.anthropic.com/index/core-views-on-ai-safety
- Anthropic releasing Claude: https://www.anthropic.com/index/introducing-claude
- Constitutional AI: https://scale.com/blog/chatgpt-vs-claude#What%20is%20%E2%80%9CConstitutional%20AI%E2%80%9D?
- Palm API opened up: https://developers.googleblog.com/2023/03/announcing-palm-api-and-makersuite.html
- Attention is all you need: https://proceedings.neurips.cc/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf
- JAC recordings: https://vimeo.com/user196160056
- Factored cognition: https://primer.ought.org/
What is ML Safety Report?
A weekly podcast updating you with the latest research in AI and machine learning safety from people such as DeepMind, Anthropic, and MIRI.