All of Jia's Comments + Replies

Effective Thesis is hiring project manager and content manager

+1 to Michael's comment. Thanks for doing what you're doing David (:

Careers Questions Open Thread

Hi Ben,

I'll be starting as an operations and research assistant at the Centre on Long-Term Risk in a few months, where I'll probably help out with AI governance topics related to multi-agent RL.

But I'm open to doing courses that are generally useful and engaging!

9Niel_Bowerman9moHey Jia, I haven't done many online courses, but one that I did and enjoyed was the Coursera Deep Learning course with Andrew Ng. https://www.coursera.org/specializations/deep-learning [https://www.coursera.org/specializations/deep-learning] I think if you will be working on multi-agent RL and haven't played around with deep learning models, you will likely find it helpful. You code up a python model that gets increasingly complicated until it does things like attempting to identify a cat (if I'm remembering it correctly). It's fairly 'hands on' but also somewhat accessible to people without a technical background. Friends of mine starting out at both CSET and OpenAI worked through it and found it helpful to get context as they moved into their new roles.
Careers Questions Open Thread

Do people have online courses to recommend?

I have ~2 months off and am considering intro courses in stats and probability, game theory, or data science. I'm open to other recommendations, of course!

This might be too elementary for you, but in college I benefited from Model Thinking by Scott Page. It's a breezy introduction to a long list of popular models used in the social and empirical sciences, and I think plausibly had a small effect on my general perspective of trying to see the world in terms of many simple models (vs eg "model-free" intuitions, or a single grand overarching model).  

6Benjamin_Todd9moHi Jia, There's a lot of options! Could you clarify which problem areas do you want to work on, and which longer-term career paths are you most interested in?
Can we convince people to work on AI safety without convincing them about AGI happening this century?
Answer by JiaNov 26, 20209

On his recent interview with FLI, Andrew Critch talks about overlaps between AI safety and current issues, and the difference between AI safety and existential safety/risk. Many (but not all) AI safety issues are relevant to current systems, so people who care about x-risks could focus on safety issues that are novel to advanced systems.

If you take a random excerpt of any page from [Aligning Superintelligence with Human Interests] and pretend that it’s about the Netflix challenge or building really good personal assistants or domestic robots, you can suc

... (read more)
Longtermist reasons to work for innovative governments

Thanks for writing this post! It's cool to see people thinking about less direct, but potentially more neglected and tractable paths to affecting influential governments.

Do you have thoughts on the difference between intentional and unintentional diffusion?

  • Intentional. Country A actively tries to export its policies. Perhaps it is trying to establish itself as a leader on a specific issue, or shape global standards in a way that benefits local companies. The literature on "niche diplomacy" might be somewhat relevant here.
  • Unintentional. A is "doing its
... (read more)
1Alexis Carlier1yHi Jia, I don't know much about policy diffusion, unfortunately. From the brief reading I have done, both seem to occur. Learning from other governments appears to be important; according to some researchers, exchange via membership in international organisations is a central causal factor. "Intentional" governments could make active engagement in these a higher priority. It's not obvious to me that status effects will be that important. Influential positions in any government are pretty well-regarded, and it could be easier to climb the policy ladder in innovative governments (e.g. if they are smaller).
AMA: Markus Anderljung (PM at GovAI, FHI)

Thanks for doing this!

People should stop using “operations” to mean “not-research”. I’m guilty of this myself, but it clumps together many different skills and traits, probably leading to people undervaluing them.

Could you say more about the different skills and traits relevant to research project management?

Thanks, Jia!

Could you say more about the different skills and traits relevant to research project management?

Understanding the research: Probably the most important factor is that you're able to understand the research. This entails knowing how it connects to adjacent questions / fields, having well thought-out models about the importance of the research. Ideally, the research manager is someone who could contribute, at least to some extent, to the research they're helping manage. This often requires a decent amount of context on the research, of... (read more)

What's the big deal about hypersonic missiles?

yup! I tried to make this point in the section on trajectory: "Hypersonic missiles fly lower than ballistic missiles, which delays detection time by ground-based radar.". I'm trying to include the following photo to illustrate the point, but I can't seem to figure out how ):

https://imgur.com/a/Ai7Ny7q

How Europe might matter for AI governance

Are you mainly referring to technical papers, or does your statement consider work from folks at FHI, CSER, CFI, etc.?

4rohinmshah2yI was excluding governance papers, because it seems like the relevant question is "will AI development happen in Europe or elsewhere", and governance papers provide ~no evidence for or against that.