I’ve been involved in Effective Altruism since 2015, including founding a society and working in movement building. Later, I quit my job as a digital consultant to work on AI. I did a data science bootcamp and the AIM Research Training Program. Now I offer freelance AI Governance research assistance and facilitate BlueDot Impact’s AI Governance course.
My Writing:
This was a great experience and I learnt a lot:
Choosing a research topic is a whole research project in itself
The writing phase takes much longer than the earlier phase of working out what to write (more than twice as long)
Co-working on research with one other person is great. It’s very motivating and you learn a lot from each other. You have faster feedback loops and so can make a better outcome sooner.
This kind of research-writing-co-working is very mentally tiring (at first I couldn’t do more than 4hrs per day)
We really wanted to complete the project in a tight timeframe. I actually posted this 2 weeks after we finished because it was the first chance I had.
Some reflections:
I think that the amount of time we set aside was too short for us, and we could still have made worthwhile improvements with more time to reflect, such as:
Choosing an easier topic for our first research project
Doing more further reading
I think the section on Risk-conducive preferences (RCPs) is not important enough to warrant the amount of words it is taking up
Many sentences could be re-written to improve the wording, and I don’t think ‘Factors of malevolent actors’ is a very good heading.
(I’ll come back and reply to this comment with more of my own reflections if I think of more and get more time in the next day or two) (edit: formatting)