2018

October

Joined the EA Hotel in October 20; joined RAISE and started catching up with their work

November

Summarized some IRL papers for RAISE (roughly the first four lessons of the course)

December

Summarized more IRL papers for RAISE (roughly the remaining four lessons of the course)

2019

January

Started reorganizing the summaries into lessons

February

Finished writing IRL Lessons

March

Helped run a RCT to test the quality of RAISE's lessons

Prepared for the coming AISC (reading papers with my team and brainstorming ideas)

April

Went to AISC and worked on a project about wireheading and embedded agency

May

(All RAISE projects got discontinued)

Wrote a paper based on the AISC project and submitted to a workshop at IJCAI

June

Paper got accepted, revised it and submitted a camera ready version

July

Read more about technical AI alignment research and AI Forecasting research

Attended the Human Aligned AI Summer School

August

Went to the IJCAI workshop and presented a poster about the wireheading paper

Assisted participants at the Learning by Doing AI Safety Workshop to help them generate solutions to AIS

Attended the Technical AI Safety Unconference

September

Self study on Data Science and Machine Learning

October

Self study on Data Science and Machine Learning

Thinking about projects for AISRP

November

Attended AISRP’s topic formation workshop, settled on a project about Newcomb-like problems for AIXI and Quasi Bayesian Agents

December

Started working on the first phase of the project

2020

January

Drafting a post that analyses AIXI in Newcomblike problems

February

Almost finished the draft, and gave feedback to Chris Leong's series of articles on the topic (see e.g. Stuck Exploration)

Became aware of a promising CV startup in Italy, evaluated whether to join them as an opportunity for earning to give, as well as for gaining experience in ML, and made some improvements to their code as a remote collaborator