Dr. David Denkenberger co-founded and is a director at the Alliance to Feed the Earth in Disasters (ALLFED.info) and donates half his income to it. He received his B.S. from Penn State in Engineering Science, his masters from Princeton in Mechanical and Aerospace Engineering, and his Ph.D. from the University of Colorado at Boulder in the Building Systems Program. His dissertation was on an expanded microchannel heat exchanger, which he patented. He is an associate professor at the University of Canterbury in mechanical engineering. He received the National Merit Scholarship, the Barry Goldwater Scholarship, the National Science Foundation Graduate Research Fellowship, is a Penn State distinguished alumnus, and is a registered professional engineer. He has authored or co-authored 156 publications (>5600 citations, >60,000 downloads, h-index = 38, most prolific author in the existential/āglobal catastrophic risk field), including the book Feeding Everyone no Matter What: Managing Food Security after Global Catastrophe. His food work has been featured in over 25 countries, over 300 articles, including Science, Vox, Business Insider, Wikipedia, Deutchlandfunk (German Public Radio online), Discovery Channel Online News, Gizmodo, Phys.org, and Science Daily. He has given interviews on 80,000 Hours podcast (here and here) and Estonian Public Radio, Radio New Zealand, WGBH Radio, Boston, and WCAI Radio on Cape Cod, USA. He has given over 80 external presentations, including ones on food at Harvard University, MIT, Princeton University, University of Cambridge, University of Oxford, Cornell University, University of California Los Angeles, Lawrence Berkeley National Lab, Sandia National Labs, Los Alamos National Lab, Imperial College, Australian National University, and University College London.
Denkenbergeršø
Thanks for doing the work to make a specific ALLFED cost effectiveness estimate! I think the AIs made a number of good points. However, I was saying the CEARCH result is ~$170 per life because GiveWell uses $5000/ālife and CEARCH was saying 30x as cost effective as GiveWell. I think GiveWell uses averting a child death means saving ~37 DALYs, and an adult death ~30 DALYs. I donāt think the AIās assumption of 80 QALYs per life saved is realistic (unless you are expecting radical life extension). The AI starts with CEARCH and then adjusts cost per life saved upward. I think there are good reasons why CEARCH is an overestimate of cost per life saved. For one, it finds nuclear risk to be significantly smaller than volcanic risk. Most analysts in this space think that the nuclear risk is significantly larger than the volcanic risk. Furthermore, the AI assumes that ALLFEDās work outside of policy for abrupt sunlight reduction scenario (ASRS) is less cost-effective than the ASRS policy work. However, I think the pandemic work is likely to be even more cost-effective, especially from the long term perspective, because pandemics are generally regarded as a greater existential risk.
The AI did seem to agree with the argument that ALLFED should have a long-term impact, it just didnāt think that the AI x-risk estimate should be used. Thatās fineāI didnāt think you would want to do a bespoke model for ALLFED, but now that you have done it for the near term, I do think it is important to do it for the long term. The AI points out that the marginal cost effectiveness calculations of the longterm impact in the journal articles are out of date because we have now spent more money. Of course thatās true, but thatās why we also calculated the cost effectiveness of spending hundreds of millions of dollars to see if the whole effort was justified. And indeed that still came out as more cost effective than AI safety. Now of course other things have changed since ~2021. AI timelines have gotten much shorter, but we were assuming that only $3 billion would be spent on AI safety, and I think itās pretty clear that a lot more than that will be spent now (especially if you count the total compensation including stock options of AI safety workers in the labs (even with your weighting of 0.3 for lab work), but that might be a topic for another post). AI 2040 hopes that trillions of dollars will be spent on AIS. Also since then, nuclear risk has gotten larger per year with the Ukraine war and potential acceleration and destabilization due to AI. Also, engineered pandemic risk per year has gone up with AI capabilities. However, this does mean a shorter number of years in expectation that the nuclear and pandemic risk might be relevant if you think the nuclear and pandemic risk will go away after AGI/āASI. For comparison, your cost per microprobability of reduction in x-risk of AI safety is $1.2 million. The median in the papers for the 3 billionth dollar on AIS was $2.5 million, with the mean being lower, so pretty good agreement with your value. So overall, since 2021, the relative marginal cost effectiveness of spending hundreds of millions of dollars on GCR resilience vs what we think will be spent on AIS I donāt think has changed too much.
The AI missed other outside evaluations of ALLFEDās longterm impact:
Speedrun: Demonstrate the ability to rapidly scale food production in the case of nuclear winter by Marie Buhl from Rethink Priorities: āmy (extremely rough) estimate that this project reduces x-risk with a cost-effectiveness of ~$260 million per 0.01% absolute reduction[1] (~70% confidence interval: 2.2 million to 2.7 billion). If this estimate were accurate, then this project would clear our median roughly estimated cost-effectiveness bar of $500M per basis-point of x-risk avertedā.
Shallow evaluations of longtermist organizations by NuƱo Sempere: āI disagree strongly with ALLFEDās estimates (probability of cost overruns, impact of ALLFEDās work if deployed, etc.), however, I feel that the case for an organization working in this area is relatively solid.ā (Note that this is a 5 year old analysis, but he recently said he respects ALLFED more now).
Iām curious if you have thoughts on how itād be if you submitted feedback like youāve done here into some sort of form, the LLMs went back and forth processing it like that, updated the site, and published a transcript like the above. I think if I make it fully automated right now itād be fairly exploitable due to LLM sycophancy, unless I tried pretty hard to mitigate that.
Interesting idea! I guess it would be less exploitable than direct edits like Wikipedia.
-
As for putting āunknownā EAs/ārationalists on the list, I see the drawback of putting a lot of them on if you are targeting a general audience. But I do think it is compelling to show that even without a long-term perspective, donating to existential risk reduction can allow everyday people to beat billionaires in terms of lives saved.
It was a significant factor, but it is also a great job.
I agree that ALLFED should not inherit the default disaster relief numbers. Ideally it would have its own analysis, because even if the numbers do end up being similar to AI x-risk, the reasons why would be pretty different.
The peer-reviewed estimates (here and here) say with high probability that ALLFED is more cost effective than AI x-risk, but external analyses were closer to on par with AI Safety. So I think itās reasonable to use the AI x-risk number, even though it is for different reasons.
By the way, I think other pandemic and nuclear work that is targeting the worst case scenarios would have some long-term impact even if the catastrophe didnāt cause extended civilizational collapse because it could make other risks more likely (e.g. global totalitarianism, or worse values ending up in AGI) or extending the time of perils.
If someone wants to add a custom effectiveness analysis of ALLFED (it looks like CEARCHās numbers are just for part of the portfolio), thatād be great. Otherwise I plan to eventually get to it.The peer-reviewed estimate (here) has higher cost effectiveness for all parts of ALLFED than the CEARCH estimate, so I think itās reasonable to use CEARCHās estimate for all of ALLFED (or just use the similar AI x-risk number). But Iād be interested in your estimates as well!
Yeah the LLM driven donation research did find both of those sources and both are used as proof for at least some donations. This wouldnāt have caused people to be added who werenāt already on the list though.
Iām confusedāthere are many people on the MIRI list who are not on your list and you said you were open to adding people who were not well known.
I think this is a useful project.
You currently have ALLFED classified as Disaster Relief. I think itās closer to AI Existential Risk because:
the main donor (Jaan Tallinn through SFF) is donating because of the long-term future effects
ALLFED does work on risks downstream of AI, including extreme pandemics
Your estimate of $137/ālife saved in 100 years for AI Existential Risk is close to CEARCHās result of ALLFED-like policy work of ~$170/ālife
As for your Longtermist calculation, it goes out 1 trillion years, but you are assuming roughly current population, so no expansion into space? Not that it matters for rankingāwhen I did longtermist cost effectiveness calculations (here and here), I just put it in terms of long-run future improvement and didnāt try to calculate lives saved.
Thanks! I havenāt thought too much about your question, because fortunately ALLFED team members are flexible with meeting times so they make my New Zealand time zone work (and really any time zone). But I would note that New Zealand is only 3 to 5 hours different from California, depending on the time of year. There are many countries outside of NATO with time zones not very far from EA hubs, but most of them are not high income and not English-speaking, so that doesnāt work for a lot of people (though itās true that living costs in lower income countries can be much lower). There are European countries outside of NATO, but I expect the EMP to still hit them. You can get some reduction in risk just living somewhat outside of big NATO cities.
I think another important consideration is risk. With AI compressing timelines, if one believes there is a 1% per year higher risk of dying due to nuclear war or extreme pandemic in a big city in NATO versus a safer place, and one values oneās life at ~$10 million, that is $100,000 per year incentive not to live there (not including family members). Furthermore, if one thinks there is some chance of radical life extension, this number would be significantly larger. This risk can be mitigated to some extent with leaving quickly, but I havenāt found many people who have such a plan.
Sounds promising. Are you interested in resilience to AI catastrophes or only prevention?
I agreeāAI estimates we could get cost parity to beef burgers by building a 100,000 tonne dry per year factory that is far larger than anything out there now (~2% of global burgers) (more like paper factory size). But that would cost close to $1B. Iām pretty sure that 2% of consumers would accept current taste of plant based burgers if they were slightly cheaper, but thatās a big investment.
Repeating my comment from LW as I would like feedback from people here too:
But perhaps instead of shutting down, an AI company could reallocate 100% of its budget on some combination of safety research + global coordination to make AI development safer, and do just those things until it runs out of money. Think of how much more safety work a they could do if they dedicated all their resources to the problem!
A lot of people think we should pause at AGI, but one problem is that AI capabilities are spiky, so we will probably get powerful and dangerous capabilities even before it is human level at all tasks. So I was thinking what might actually be politically/āeconomically feasible is pausing further development once they get an automated AI researcher, because all those AI capability researchers are not going to be happy about being laid off. So we could turn them into AI safety researchers!
Very helpful!
Very quickly cargo ships would be queuing up for days and even weeks as ports are not able to process their containers. From this perspective, the 1 to 2 days of navigation delay due to the loss of GPS becomes insignificant because it does not matter if a ship is a day late arriving at a port if it will need to queue for a week before being unloaded.
Why would losing GPS prevent processing?
But $1.8 trillion (about $5B/āday for a year) for COVID for US economy seems like a good rough estimate.
Even COVID can be considered an existential risk if it prolongs the time of perils. Whatās your estimate of the global economic impact over a year?
AI estimates that global knowledge work is ~$35T/āyr and that the value of AI is ~6x the revenue (though that could change). That would imply we would only need ~$6T/āyr AI revenue to substitute for all knowledge work. Now it is true that AI would be affecting the capital side of gross world product (GWP) (~$60T/āyr), and that the GWP would grow by then. But there is the time delay of diffusion, indicating that the impact of the AI at the time would be much larger eventually. So I think even $1-3T/āyr revenue with strong growth would be fairly strong evidence for AGI/āTAI.
Yeah, the funding situation in nuclear security is still dire. Depending on how you count, the cuts in US government funding various things that charities may want to make up for were greater than $50 billion per year.
I agree there is some overlap between these principles and EA principles. But I still think the EAs are more willing to debate them. For instance, the first implies torture is never justified (and this is a common view in UU), but some EAs will debate the cases where it could be justified. Also, maximizing utility does not necessarily push you to justice and equity. Spiritual growth does not apply to most of EA. And EAs generally question environmental/ābiodiversity principles.
Global panĀdemic preĀvenĀtion: how disĀease spreads across borĀders and the three counĀtries deĀterĀminĀing panĀdemic outcomes
To be clear, my relocation post was for getting ready to relocate quickly in response to a trigger such as a new infectious, fatal disease being discovered.
No, I donāt think mirror bio is coming sooner.
I would like to know why you donāt think it would make sense to relocate to a place like New Zealand or Australia. My thought is that they demonstrated that they had the will and the functioning government to suppress COVID, and can take advantage of their isolation. You may be able to prevent your family from getting the disease in the US, but then you may have to deal with no electricity or water or fuel if the vital employees are unable or unwilling to show up to work. Australia could likely keep industry functioning conventionally, but even New Zealand might be able to improvise biofuels or go back to animal power.
But EA seems increasingly ideological to me, it seems to be a lot less concerned with asking and answering a genuinely open question, and a lot more concerned with causing people to follow established paths to impact.
That may be true, though I would note that EA should have been more open to new causes in the beginning, until it did a bunch of research and found promising cause areas. I will also note that there is evidence that EA is still open to new cause areas, such as the recent shift from focusing on AI catastrophes to making AI futures go well (not to mention many individual EAs shifting between cause areas within EA).
But if EA were more ideological, does this mean it is dying? Maybe it is a worse fit for some people as a community. But I think the most important question is, āIs EA having more positive impact?ā I agree that that is not just a function of jobs switched and money moved. But overall, I think it is having more positive impact, so I would not call that dying.
As in my other comment, I am mixing deference to orthodoxy and leaders. I have similar experience duration in academia. I think questioning things like political left positions and the value of education is more discouraged in academia than questioning the common beliefs in EA. I think it also depends on the field. In some academic fields, there are specific camps with a lot of tribal behavior. Thereās a quote that science advances one funeral at a time. I think that EAs are more open to changing their opinions. There is also a lot of tribal behavior between fields in academiaāI think there is significantly more bias towards oneās strand of academia than towards oneās strand of EA. Maybe thatās not deference to leaders, but the questioning of the value of oneās field in academia is uncommon. Whereas in EA, there is lots of questioning whether different strands are net negative.
However, I do agree that some EAs like taking contrary opinions and might defer to a random blogger against scientific consensus. Those EAs often have a low opinion of peer review, and are probably more often wrong than the scientific consensus.
Iām blending deference to leaders and orthodoxy here. In UU, questioning of traditional theology is encouraged; questioning the views of the political left, much less so. Of course EA has its own orthodoxy, but questioning and criticism of it is much more encouraged. I think there has been more criticism of Moskovitz, Karnofsky, MacAskill, and Ord in EA than typical leaders in UU (e.g. ministers), but maybe your experience is different?
Thanks!