Karen, your reply appears to be AI-generated. I recommend avoiding using LLMs in job applications and EA Forum conversations -- the prose isn't good and (most) people still prefer talking with humans.

Great summary! I personally like #3 Research x Military the most. This approach seems very useful to the EA community, especially those working on AI x national security, biosecurity, nuclear risks, or anything that could possibly have a military element (e.g., wargaming, conflict risks due to climate change). Maybe it could evolve into something like a military-oriented Future Impact Group? 

On this direction, you mention a couple possible drawbacks. First: 

It's obvious why this would be valuable to the researchers, but I'm not sure how to make this worthwhile for the military people. I'd like to think that being "friendly forces" for the UK military will go a long way, but ultimately, I would be cashing in credit with these people. Neither the researchers nor I can meaningfully solve their problems. Perhaps this needs to be combined with a pitch that I could help connect them with high-impact civilian careers down the line, as described in #2. Alternatively, I could include a request for soft commitments to help out with research stuff in the advising strategies discussed above. If they found my career advice helpful, I expect most people would be happy to give their time in return.

I think you may be underestimating the value for military folks. If they're actually interested in EA, then supporting EA research is a great opportunity to learn how top orgs work and how staff think and write. It might also help them get a job down the line, but I'd avoid making any promises. 

Second: 

In some fields, much of the expertise that could be helpful to researchers is classified. Furthermore, there are limitations in what people from one country are allowed to say to people from another. 

This is a fair point, but in my experience, EA orgs focus more on strategic and high operational-level issues on which there is plenty of non-classified material. It's less "What are typical UK submarine routes?" and more "How does the UK currently detect submarines, and how might we expect advanced AI to improve those capabilities?"

Quick thoughts on the other potential directions for HI-DEF: 

#2 Targeted advising/outreach -- This seems useful given your experience and network. Maybe you could focus on recruiting talented OF2/OF3?

#4 Community engagement -- I can see the value add here. But hosting a newsletter, Discord, or occasional meet-ups feels more like a complement to #3 or #2 than a substitute.

#1 General advising -- I'm less excited about this option. As you note, there are several orgs that already do this effectively. But I do like your idea to partner with these orgs to serve as a possible point of contact for military folks. 

A social scene where everyone hangs out is easier to be a part of, but then - whats the point? Plus, lots of people already have a roster full of friends and families and don't need another social group. 

I strongly agree with this point. I enjoy hanging out with EA friends, but if our objective shifts from "doing good better" to having fun, then I'd probably be less involved with the community.

This is great! Props to Austin and Marcus for starting this conversation, and you (and Chana) for expanding on it. 

As Austin puts it, there are actually at least three ways an AI Safety video could have an impact:

1. Introducing people to a topic
2. Convincing people to take an action (change careers, donate)
3. Providing information to people working in the field

I think this is right, and as you say, it's hard to compare/rank videos when they have such different objectives. We don't compare blogs to books, so comparing TikToks with high-production videos seems to be wrong for the same reasons.

Thanks for sharing. I agree that it's useful to know how "thought leaders" interpret EA.

It's disappointing, however, that Mounk's primary critiques of EA are: 

• An apocryphal story about some asshole taking EVs to the extreme.
• SBF, a fraud who has no legitimate EA backers today, and, like the imaginary dude above, took EVs to the extreme.
• Worries that EAs earning to give might end up donating less than planned or—God forbid—roughly the same amount as everyone else.

Mounk's critique of longtermism is okay, but Kelsey Piper raised the same objections but better two years ago.

Good point. I prefer some ambiguity over a longer title, but welcome alternatives that are both clear and concise.

I look forward to reading your longer report. If you're interested in convincing skeptics like me, I suggest addressing the following questions:

Can Muslims give Zakat to EA causes? We agree that Zakat and longtermist projects aren't compatible, and that direct cash transfers to the poor are compatible. But what about other cause areas? I'm not familiar with Zakat being used for non-emergency global health interventions (e.g., deworming, reducing lead exposure, deploying insecticide-treated bed nets, etc.). It seems possible, given Zakat use for medical necessities during humanitarian crises (e.g., Gaza), but are there examples that you can provide here outside the context of war or natural disasters? If none currently exist, are there any orthodox Muslim scholars who support this use of Zakat?

What are the specific inefficiencies with Islamic Relief, Helping Hand, and other Muslim-run charities that concern you? You inspired me to do some research, and I've found some worrying reports about Islamic Relief USA "double dipping" on administrative fees, and LaunchGood not actually verifying their "Zakat-verified" campaigns. But generally speaking, do you think the top Muslim charities have higher administrative costs than those of other charities? If so, what's your evidence?

Relatedly: Do you know if GiveDirectly considered their Zakat fund for Yemen a success? Given that 100% of the funds collected were distributed to Yemeni families in need, is this kind of program sustainable?

How will you convince Muslims to redirect their Zakat to non-Muslim charities? I suspect that a significant number of Muslims (including myself) will hesitate to do so. We're fine giving Sadaqah to non-Islamic institutions, but there's a general belief that Zakat should be distributed by, with, and through fellow Muslims. How do you propose tackling that issue?

How will you counter local-first doctrine? We're often instructed that Zakat should be "focused locally to bring about change in the very environment in which the Zakat payers live." (Yaqeen, p. 13) I personally don't follow this rule, but it's a popular one, and it pushes people to give Zakat to their local Masjid instead of internationally, where they could get the biggest bang for their buck. This principle does not appear to apply to Sadaqah, as far as I know, which is a contributing factor for why I think it's easier to pull Muslims into GWWC than to shift their Zakat practices.

Lastly, one response: 

It's not clear yet whether or not the best path to impact would be to first e.g. a) get existing effective orgs to become zakat-compliant, b) get already-zakat-compliant orgs to become more cost effective, c) create a zakat-complaint org which collects donations and distributes them to cost effective orgs, d) create an org which only convinces and promotes effective zakat and then directs people to those orgs without collecting zakat ourselves. 

I don't think this is true. If you can do (B)—convince Zakat-compliant organizations to become more cost effective—so long as their Zakat programming is mostly direct cash transfers, that would be massive impact with minimal effort. You wouldn't need to start a new org and staff it and get it Zakat-certified, or get an existing EA organization Zakat-certified and convince Muslims to trust them.

I appreciate how long you've been working on this topic! I've read your related posts and the conversations are fascinating. 

That said, I don't think that plugging Zakat into an EA framework is possible. I haven't seen any serious Muslim scholar support using Zakat funds to avoid global catastrophic risks or future pandemics. You've already been Fatwa shopping to no avail. 

There's only one clear area of overlap: direct cash transfers to poor Muslims. But I'm unsure how your group would conduct this work more effectively than popular charities like Islamic Relief or Helping Hand. 

I have two alternative recommendations for your proposed group (and for any Muslim in EA): 

1. Invite Muslims to EA: We can advise Muslims to donate to EA charities for the same reason that others do: to do the most good. It won't count towards their obligatory Zakat, but it will count as voluntary Sadaqah. We ought to convince Muslims that giving 2.5% annually is a very low baseline and we should do more! 
2. Lobby Muslim charities to be better: You say there's no accountability / transparency demonstrated by the most popular Zakat-collection and distribution groups. Perhaps your group can write on what specific action / information is missing, why this is a problem worth addressing, and how these charities can do so. 

This is a great post. Your stories on Jayantha Dhanapala, Franklin Miller, Roberta Wohlstetter, and Joan Rohlfing are particularly helpful in illustrating unique theories of change. 

On "What goals should we be aiming towards?" I recommend that people enter the field without specific goals until they better understand the field and its recurring debates. Focusing on a broader aim -- like reducing the risks and consequences of nuclear war -- is great because it keeps you mentally flexible. You're not tied down to the preferences of the primacist, arms control, or disarmament communities. 

This is important because nuclear experts often disagree on what polices are stabilizing or destabilizing. For example, in that same section, you write:  

Prevent deployment of the most escalatory weapons. For example, in early 2023, the US decided not to develop a new nuclear-armed sea-launched cruise missile (SLCM-N). SLCM-Ns are considered destabilising because of their target and payload ambiguity, so this decision was a substantial success for risk-reduction. Similarly, there may be policies that could disincentivize Russia and China from deploying some of the destabilising weapons they are currently pursuing.

Yet Rob Soofer, Heather Williams, and other nuclear scholars would argue that the SLCM-N is stabilizing because it deters Russian use of low-yield (aka "non-strategic") nuclear weapon. Satisfying that mission may outweigh the possible drawbacks on target and payload ambiguity.