This repository documents my (totally incomplete) exploration of the philosophy of ethics and society with a particular interest in the future role of artificial intelience. This will include both near-term and long-term future issues. The content is based on my reading, research, thought, coursework,
This repository is a perpetual work in progress. If we're lucky, one day it might become a website, or even a book (haha, sure). But seriously, if you want specific resources or progress or if you have questions, just raise an issue.
-
[WIP] Resources (below): A collection of resources and links related to the topics of this project. Includes notable readings, other reading and resource lists, etc.
-
COMP90087 The Ethics of Artificial Intelligence: Unimelb's COMP90087 is an introduction to current issues in AI and society, framed using introductory moral philosophy. I joined the teaching team in the innaugural semester (2021). Here you can find some information about the subject and staff, the official syllabus, and a complete list of readings!
-
[WIP] AGI Safety Fundamentals: A fellowship (more of a reading group) from EA Cambridge, on research into the safety of generally intelligent AI systems. Here, find readings, and possibly some of my notes if I decide to take some throughout the program. The readings are already available online now: Fellowship on AGI Fundamentals.
-
[WIP] My notes on ethics and AI: My list of key topics, lessons, and readings, and some of my remaining questions.
So far completely unsorted. I will progressively organise and arrange these.
Wikipedia has some very broad, historical paper lists (with summaries):
- List of important publications in computer science / AI [wiki]
- List of important publications in theoretical computer science [wiki]
Yoni Nazarathy's course had a list of papers important to deep learning specifically:
- Key papers in the development of deep learning 1958--2017 [deeplearningmath.org] (more references below)
CHAI and Berkeley has some resource lists:
-
CHAI annotated bilbiography [humancompatible.ai]
-
Reading list for CS 294-149: Safety and Control for Artificial General Intelligence (Fall 2018) [course page]
Other bibliographic sources:
- The Alignment Newsletter Database [gsheet] (all previous entries in Rohin Shah's Alignment Newsletter)
- TAI Safety Bibliographic Database (2016--2020) [alignment forum] (aims to be comprehensive, including some of the other lists on this page; also contains lists of organisations and some reviews)
Wikipedia articles of interest
- AI control problem https://en.wikipedia.org/wiki/AI_control_problem
AI Safety Support
- links page https://www.aisafetysupport.org/resources/lots-of-links
- wishlist/bottleneck survey https://forum.effectivealtruism.org/posts/2pxGXYX2JrptvLpzZ/ai-safety-career-bottlenecks-survey-responses-responses (part career stuff, part resources and reading lists)
MIRI
- motivations https://intelligence.org/why-ai-safety/
- research guide (big textbook and essay/paper reading list) https://intelligence.org/research-guide/
Fora
- Alignment forum https://www.alignmentforum.org/
- Check the curated sequences and community sequences
- There's also an archive of old AI internet posts by Richard Ngo
- On the topic of Richard Ngo this might be worth a read https://www.lesswrong.com/posts/k6NkvAcRaKBMAzqEF/my-intellectual-influences
Other
-
https://www.cser.ac.uk/research/risks-from-artificial-intelligence/
-
https://www.cnas.org/artificial-intelligence-and-global-security-reading-list
-
Apparently there's an active, open, AI safety slack (see AGISF slack for link)
-
The alignment newsletter is pretty great (especially the podcast)
-
Chris Olah on transparency for AI safety (see alignment newsletter 72)
-
https://docs.google.com/document/d/1FbTuRvC4TFWzGYerTKpBU7FJlyvjeOvVYF2uYNFSlOc/edit
-
artibal.com/explore/ai_alignment
-
Causal influence diagrams https://www.lesswrong.com/posts/Cd7Hw492RqooYgQAS
-
Natural abstraction https://www.alignmentforum.org/posts/cy3BhHrGinZCp3LXE/testing-the-natural-abstraction-hypothesis-project-intro
-
https://www.cold-takes.com/roadmap-for-the-most-important-century-series/ blog and seris for another overview of it all
-
sutton's website has a list of some key papers http://www.incompleteideas.net/publications.html