Current plan as of July 10, 2023

Goal

My goal is to orient in the field of AI safety. I want good enough models of the scenarios, risks, state of the field, and personal fit, than I can generate a small list of high impact actions I could take and their theory of change.

I’m looking to generate actions like “train in this domain to get this job in this org”, or “start this org to develop this product to solve this problem to reduce this risk”, or “research this topic to understand this risk to plan how to avoid it in this scenario”.

Once I have this list, I will move to a phase of trying my fit in each of them.

Plan

First phase (1 week)

The goal is to improve the second phase plan by doing more preliminary research:

• How others recommend getting into the field
• Stories and lessons learned of other people who managed or not to get into the field
• How to develop expertise in a field in general

Gather techniques and advices from this research, and add details to my initial day-to-day process in the second phase.

Get advice from lots of people in different parts of AI safety on this plan

Find advisors, mentors, and people to bounce ideas with and do regular accountability checkups.

Second phase (2-3 month)

The goal is to build my models starting with those four components

1. What are the different harm scenarios ? What are their current probability if I do nothing ?
2. What are the core difficulties leading to harm ? What could solve them ? Which actions influence them ?
3. What are all the current plans ? Who is working on which part of which plan ?
4. What can I work on ? What is my comparative advantage ? What would motivate me ?