A new paper: “Categorizing Variants of Goodhart’s Law”
New research write-ups and discussions: Distributed Cooperation; Quantilal Control for Finite Markov Decision Processes
New at AI Impacts: Transmitting Fibers in the Brain: Total Length and Distribution of Lengths
Scott Garrabrant, the research lead for MIRI’s agent foundations program, outlines focus areas and 2018 predictions for MIRI’s research.
Scott presented on logical induction at the joint Applied Theory Workshop / Workshop in Economic Theory.
Nautilus interviews MIRI Executive Director Nate Soares.
From Abram Demski: An Untrollable Mathematician Illustrated
News and links
From FHI’s Jeffrey Ding: “Deciphering China’s AI Dream.”
OpenAI researcher Paul Christiano writes on universality and security amplification and an unaligned benchmark. Ajeya Cotra summarizes Christiano’s general approach to alignment in Iterated Distillation and Amplification.
Christiano discusses reasoning in cases “where it’s hard to settle disputes with either formal argument or experimentation (or a combination), like policy or futurism.”
From Chris Olah and collaborators at Google and CMU: The Building Blocks of Interpretability.
From Nichol, Achiam, and Schulman at OpenAI: Reptile: A Scalable Meta-Learning Algorithm.
The post April 2018 Newsletter appeared first on Machine Intelligence Research Institute.