Scott Garrabrant has concluded the main section of his Finite Factored Sets sequence (“Details and Proofs”) with posts on inferring time and applications, future work, and speculation.
Scott’s new frameworks are also now available as a pair of arXiv papers: “Cartesian Frames” (adapted from the Cartesian Frames sequence for a philosopher audience by Daniel Hermann and Josiah Lopez-Wild) and “Temporal Inference with Finite Factored Sets” (essentially identical to the “Details and Proofs” section of Scott’s sequence).
Other MIRI updates
- DeepMind’s Rohin Shah has written his own introduction to finite factored sets.
- Alex Appel extends the idea of finite factored sets to countable-dimensional factored spaces.
- Open Philanthropy’s Joe Carlsmith has written what’s probably the best existing introduction to MIRI-cluster work on decision theory: Can You Control the Past?. See also Carlsmith’s decision theory conversation with MIRI’s Abram Demski and Scott Garrabrant.
- From social media: Eliezer Yudkowsky discusses paths to AGI and the ignorance argument for long timelines, and talks with Vitalik Buterin about GPT-3 and pivotal acts.
News and links
- A solid new introductory resource: Holden Karnofsky has written a series of essays on his new blog (Cold Takes) arguing that “the 21st century could be the most important century ever for humanity, via the development of advanced AI systems that could dramatically speed up scientific and technological advancement”. See also Holden’s conversation with Rob Wiblin on the 80,000 Hours Podcast.
- The Future of Life Institute announces the Vitalik Buterin PhD Fellowship in AI Existential Safety, “targeted at students applying to start their PhD in 2022”. You can apply at https://grants.futureoflife.org/; the deadline is Nov. 5.
- OpenAI releases Codex, “a GPT language model fine-tuned on publicly available code from GitHub”.
