[Question] How should technical AI researchers best transition into AI governance and policy?

[GabeM]

In the spirit of Career Conversations Week, throwing out some quick questions that I hope are also relevant for others in a similar position!

I’m an early-career person aiming to have a positive impact on AI safety. For a couple of years, I’ve been building skills towards a career in technical AI safety research, such as:

• Publishing ML safety research projects

• Doing a Master’s degree in machine learning at a top ML school

• Generally focusing on building technical ML research skills and experience at the expense of other forms of career capital.

However, I’m now much more strongly considering paths to impact that route through AI governance, including AI policy, than pure technical alignment research. Since I still feel pretty junior, I think I have room to explore a bit. However, I’m not junior enough to have a fresh degree in front of me (e.g. to choose to study public policy), and I feel like I have a strong fit for technical ML skills and knowledge, including explaining technical concepts to non-technical audiences, that I want to leverage.

What are some of the best ways for people like me to transition from technical AI safety research roles into more explicit AI governance and policy? So far, I’m only really aware of:

• Policy fellowships that might take technical researchers without policy experience, like the Horizon, STPI, PMF, or STPF fellowships

• Policy positions in top AI labs, which are themselves important for AI governance and could transition well into other AI governance careers

• Policy research positions that require significant technical knowledge at organizations like GovAI

• Some vague notion of “being a trusted scientific advisor to key decision makers in DC or London,” though I’m not sure what this practically looks like or how to get there.

Any other ideas? Or for those who have been in a similar situation, how have you thought about this?

This post is part of the September 2023 Career Conversations Week. You can see other Career Conversations Week posts here.

[US Policy Careers]

Here are some different options for people with technical backgrounds to pivot into policy careers:

• Policy fellowships (see our database here) are a great entryway into DC policy careers for technical people (and for many non-technical people!). Fellowships are especially helpful for mid-career technical folks who would otherwise struggle to make the pivot because they’re both (1) too senior for the normal entryways (e.g. policy internships [such as in Congress], a policy-oriented (under)graduate degree, junior policy jobs), and (2) have too little policy experience to qualify for mid-level or senior policy jobs. There are policy fellowships for people from all experience levels.

  • Here are several additional fellowships which have a preference for technical people: TechCongress, Mirzayan Fellowship, Presidential Innovation Fellowship, US Digital Corps, Epidemic Intelligence Service (mainly for MDs), National Nuclear Security Administration (NNSA) Graduate Fellowship Program (NGFP), DARPA innovation fellowship, X-Force Fellowship (DOD), SMART Scholarship Program (DOD), CyberCorps Scholarship for Service (SFS), etc.

  • And there are plenty of great fellowships that, while not specifically targeting technical people, are open to them, such as PMF.

• Check out our advice on policy internships (the linked post targets undergraduates, but the internship advice applies more widely), which are the most common way for junior people to test their fit for policy and build policy-relevant career capital, whether they have technical or non-technical backgrounds.

  • You might also conduct a policy-relevant research project during a summer/​winter research fellowships offered by organizations like GovAI, ERA, CHERI, SERI, and XLab.

• If you’re currently enrolled in a technical (under)graduate degree, try to gain some policy-relevant knowledge, networks, and skills, via choosing some policy classes if you can, especially in science and technology policy, or choosing a policy-relevant thesis project.

• Participate in policy-relevant online programs, like the AI Safety Fundamentals Course’s Governance Track, speaker series like this, or these AI policy and biosecurity policy online courses.

• Consider doing a policy-relevant graduate degree, particularly a policy master’s or law degree. You can often get into these degree programs even if you have only done technical work in the past (ideally, you should be able to tell a narrative about how your interest in policy work is connected to your prior technical studies and/​or work experience). Even if you already have a technical graduate degree, it might make sense to do another (short/​part-time) policy degree if you’re serious about pivoting into policy but are otherwise struggling to make the switch.

One brief comment on mindset: Policy jobs typically don’t require people to have a particular subject background, though there are exceptions. There are plenty of people with STEM degrees and technical work experience who have pivoted into policy roles, often focused on science and technology (S&T) policy areas, where they can leverage their technical expertise for added credibility and impact. There are certain policy roles and institutions that prefer people with technical backgrounds, such as many roles in the White House OSTP, NSF, DOE, NIH, etc. So, you shouldn’t feel like it’s impossible to pivot from technical to policy work, and there are resources to help you with this pivot. We particularly recommend speaking with an 80,000 Hours career adviser about this.

[GabeM]

This is sublime, thank you!

[Zach Stein-Perlman]

(Mostly I don’t know.)

On policy fellowships: also RAND TASP.

I think many reasonably important policy roles don’t require policy experience—working for key congressional committees or federal agencies.

[GabeM]

Reposting an anonymous addition from someone who works in policy:

Your list of options mostly matches how I think about this. I would add:

• Based on several anecdotal examples, the main paths I’m aware of for becoming a trusted technical advisor are “start with a relevant job like a policy fellowship, a job doing technical research that informs policy, or a non-policy technical job, and gradually earn a reputation for being a helpful expert.” To earn that reputation, some things you can do are: become one of the people who knows most about some niche but important area (anecdotally, “just” a few years of learning can be sufficient for someone to become a top expert in areas such as compute governance or high-skill immigration policy, since these are areas where no one has decades or experience — though there are also generalists who serve as trusted technical advisors); taking opportunities that come your way to advise policymakers (such opportunities can be common once you have your first policy job, or if you can draw on a strong network while doing primarily non-policy technical work); and generally being nice and respecting confidentiality. You don’t need to be a US citizen for doing this in the US context.

• In addition to GovAI, other orgs where people can do technical research for AI policy include:

  • RAND and Epoch AI

  • Academia (e.g. I think the AI policy paper “What does it take to catch a Chinchilla?” was written as part of the author’s PhD work)

  • AI labs

[Yadav]

Hmm I’d very keen to see what an answer to this might look like. I know some people I work with are interested in making a similar kind of switch.