The very first sentence on that page reads (emphasis mine):
The Long-Term Future Fund aims to positively influence the long-term trajectory of civilization by making grants that address global catastrophic risks, especially potential risks from advanced artificial intelligence and pandemics.
I personally think that’s quite explicit about the focus of the LTFF, and am not sure how to improve it further. Perhaps you think we shouldn’t mention pandemics in that sentence? Perhaps you think “especially” is not strong enough?
An important reason why we don’t make more grants to prevent pandemics is that we only get few applications in that area. The page serves a dual purpose: it informs both applicants and donors. Emphasizing pandemics less could be good for donor transparency, but might further reduce the number of biorisk-related applications we receive. As Adam mentions here, he’s equally excited about AI safety and biosecurity at the margins, and I personally mostly agree with him on this.
Here’s a spreadsheet with all EA Funds grants (though without categorization). I agree a proper grants database would be good to set up at some point; I have now added this to my list of things we might work on in 2021.
We prioritize AI roughly for the reasons that have been elaborated on at length by others in the EA community (see, e.g., Open Phil’s report), plus additional considerations regarding our comparative advantage. I agree it would be good to provide more transparency regarding high-level prioritization decisions; I personally would find it a good idea if each Fund communicated its overall strategy for the next two years, though this takes a lot of time. I hope we will have the resources to do this sometime soon.
I personally think that’s quite explicit about the focus of the LTFF, and am not sure how to improve it further. Perhaps you think we shouldn’t mention pandemics in that sentence? Perhaps you think “especially” is not strong enough?
I don’t think it’s appropriate to discuss pandemics in that first sentence. You’re saying the fund makes grants that “especially” address pandemics, and that doesn’t seem accurate. I looked at your spreadsheet (thank you!) and tried to do a quick classification. As best I can tell, AI has gotten over half the money the LTFF has granted, ~19x the amount granted to pandemics (5 grants for $114,000). Forecasting projects have received 2.5x as much money as pandemics, and rationality training has received >4x as much money. So historically, pandemics aren’t even that high among non-AI priorities.
If pandemics will be on equal footing with AI going forward, then that first sentence would be okay. But if that’s the plan, why is the management team skillset so heavily tilted toward AI?
An important reason why we don’t make more grants to prevent pandemics is that we only get few applications in that area. The page serves a dual purpose: it informs both applicants and donors. Emphasizing pandemics less could be good for donor transparency, but might further reduce the number of biorisk-related applications we receive. As Adam mentions here, he’s equally excited about AI safety and biosecurity at the margins, and I personally mostly agree with him on this.
I’m glad there’s interest in funding more biosecurity work going forward. I’m pretty skeptical that relying on applications is an effective way to source biosecurity proposals though, since relatively few EAs work in that area (at least compared to AI) and big biosecurity funding opportunities (like Open Phil grantees Johns Hopkins Center for Health Security and Blue Ribbon Study Panel on Biodefense) probably aren’t going to be applying for LTFF grants.
Regarding the page’s dual purpose, I’d say informing donors is much more important than informing applicants: it’s a bad look to misinform people who are investing money based on your information.
We prioritize AI roughly for the reasons that have been elaborated on at length by others in the EA community (see, e.g., Open Phil’s report), plus additional considerations regarding our comparative advantage. I agree it would be good to provide more transparency regarding high-level prioritization decisions; I personally would find it a good idea if each Fund communicated its overall strategy for the next two years, though this takes a lot of time. I hope we will have the resources to do this sometime soon.
There’s been plenty of discussion (including that Open Phil report) on why AI is a priority, but there’s been very little explicit discussion of why AI should be prioritized relative to other causes like biosecurity.
Open Phil prioritizes both AI and biosecurity. For every dollar Open Phil has spent on biosecurity, it’s spent ~$1.50 on AI. If the LTFF had a similar proportion, I’d say the fund page’s messaging would be fine. But for every dollar LTFF has spent on biosecurity, it’s spent ~$19 on AI. That degree of concentration warrants an explicit explanation, and shouldn’t be obscured by the fund’s messaging.
Thanks, I appreciate the detailed response, and agree with many of the points you made. I don’t have the time to engage much more (and can’t share everything), but we’re working on improving several of these things.
Thanks Jonas, glad to hear there are some related improvements in the works For whatever it’s worth, here’s an example of messaging that I think accurately captures what the fund has done, what it’s likely to do in the near term, and what it would ideally like to do:
The Long-Term Future Fund aims to positively influence the long-term trajectory of civilization by making grants that address global catastrophic risks or promote the adoption of longtermist thinking. While many grants so far have prioritized projects addressing risks posed by artificial intelligence (and the grantmakers expect to continue this at least in the short term), the Fund is open to funding, and welcomes applications from, a broader range of activities related to the long-term future.
I personally think that’s quite explicit about the focus of the LTFF, and am not sure how to improve it further. Perhaps you think we shouldn’t mention pandemics in that sentence? Perhaps you think “especially” is not strong enough?
I agree with you that that’s pretty clear. Perhaps you could just have another sentence explaining that most grants historically have been AI-related because that’s where you receive most of your applications?
On another note, I can’t help but feel that “Global Catastrophic Risk Fund” would be a better name than “Long-term Future Fund”. This is because there are other ways to improve the long-term trajectory of civilisation than by mitigating global catastrophic risks. Also, if you were to make this change, it may help distinguish the fund from the long-term investment fund that Founders Pledge may set up.
Some of the LTFF grants (forecasting, long-term institutions, etc.) are broader than GCRs, and my guess is that at least some Fund managers are pretty excited about trajectory changes, so I’d personally think the current name seems more accurate.
Ah OK. The description below does make it sound like it’s only global catastrophic risks.
The Long-Term Future Fund aims to positively influence the long-term trajectory of civilization by making grants that address global catastrophic risks, especially potential risks from advanced artificial intelligence and pandemics.
Perhaps include the word ‘predominantly’ before the word “making”?
The second sentence on that page (i.e. the sentence right after this one) reads:
In addition, we seek to promote, implement, and advocate for longtermist ideas, and to otherwise increase the likelihood that future generations will flourish.
“Predominantly” would seem redundant with “in addition”, so I’d prefer leaving it as-is.
Which of these two sentences, both from the fund page, do you think describes the fund more accurately?
The Long-Term Future Fund aims to positively influence the long-term trajectory of civilization by making grants that address global catastrophic risks, especially potential risks from advanced artificial intelligence and pandemics. (First sentence of fund page.)
Grants so far have prioritized projects addressing risks posed by artificial intelligence, and the grantmakers expect to continue this at least in the short term. (Located 1500 words into fund page.)
I’d say 2 is clearly more accurate, and I think the feedback you’ve received about donors being surprised at how many AI grants were made suggests I’m not alone.
Could you operationalize “more accurately” a bit more? Both sentences match my impression of the fund. The first is more informative as to what our aims are, the second is more informative as to the details of our historical (and immediate future) grant composition.
My sense is that the first will give people an accurate predictive model of the LTFF in a wider range of scenarios. For example, if next round we happen to receive an amazing application for a new biosecurity org, the majority of the round’s funding could go on that. The first sentence would predict this, the second not.
But the second will give most people better predictions in a “business as usual” case, where our applications in future rounds are similar to those of current rounds.
My hunch is that knowing what our aims are is more important for most donors. In particular, many people reading this for the first time will be choosing between the LTFF and one of the other EA Funds, which focus on completely different cause areas. The high-level motivation seems more salient than our current grant composition for this purpose.
Ideally of course we’d communicate both. I’ll think about if we should add some kind of high-level summary of % of grants to different areas under the “Grantmaking and Impact” section which occurs earlier. My main worry is this kind of thing is hard to keep up to date, and as described above could end up misleading donors in the other direction, if our application pool suddenly changes.
The very first sentence on that page reads (emphasis mine):
I personally think that’s quite explicit about the focus of the LTFF, and am not sure how to improve it further. Perhaps you think we shouldn’t mention pandemics in that sentence? Perhaps you think “especially” is not strong enough?
An important reason why we don’t make more grants to prevent pandemics is that we only get few applications in that area. The page serves a dual purpose: it informs both applicants and donors. Emphasizing pandemics less could be good for donor transparency, but might further reduce the number of biorisk-related applications we receive. As Adam mentions here, he’s equally excited about AI safety and biosecurity at the margins, and I personally mostly agree with him on this.
Here’s a spreadsheet with all EA Funds grants (though without categorization). I agree a proper grants database would be good to set up at some point; I have now added this to my list of things we might work on in 2021.
We prioritize AI roughly for the reasons that have been elaborated on at length by others in the EA community (see, e.g., Open Phil’s report), plus additional considerations regarding our comparative advantage. I agree it would be good to provide more transparency regarding high-level prioritization decisions; I personally would find it a good idea if each Fund communicated its overall strategy for the next two years, though this takes a lot of time. I hope we will have the resources to do this sometime soon.
I don’t think it’s appropriate to discuss pandemics in that first sentence. You’re saying the fund makes grants that “especially” address pandemics, and that doesn’t seem accurate. I looked at your spreadsheet (thank you!) and tried to do a quick classification. As best I can tell, AI has gotten over half the money the LTFF has granted, ~19x the amount granted to pandemics (5 grants for $114,000). Forecasting projects have received 2.5x as much money as pandemics, and rationality training has received >4x as much money. So historically, pandemics aren’t even that high among non-AI priorities.
If pandemics will be on equal footing with AI going forward, then that first sentence would be okay. But if that’s the plan, why is the management team skillset so heavily tilted toward AI?
I’m glad there’s interest in funding more biosecurity work going forward. I’m pretty skeptical that relying on applications is an effective way to source biosecurity proposals though, since relatively few EAs work in that area (at least compared to AI) and big biosecurity funding opportunities (like Open Phil grantees Johns Hopkins Center for Health Security and Blue Ribbon Study Panel on Biodefense) probably aren’t going to be applying for LTFF grants.
Regarding the page’s dual purpose, I’d say informing donors is much more important than informing applicants: it’s a bad look to misinform people who are investing money based on your information.
There’s been plenty of discussion (including that Open Phil report) on why AI is a priority, but there’s been very little explicit discussion of why AI should be prioritized relative to other causes like biosecurity.
Open Phil prioritizes both AI and biosecurity. For every dollar Open Phil has spent on biosecurity, it’s spent ~$1.50 on AI. If the LTFF had a similar proportion, I’d say the fund page’s messaging would be fine. But for every dollar LTFF has spent on biosecurity, it’s spent ~$19 on AI. That degree of concentration warrants an explicit explanation, and shouldn’t be obscured by the fund’s messaging.
Thanks, I appreciate the detailed response, and agree with many of the points you made. I don’t have the time to engage much more (and can’t share everything), but we’re working on improving several of these things.
Thanks Jonas, glad to hear there are some related improvements in the works For whatever it’s worth, here’s an example of messaging that I think accurately captures what the fund has done, what it’s likely to do in the near term, and what it would ideally like to do:
The Long-Term Future Fund aims to positively influence the long-term trajectory of civilization by making grants that address global catastrophic risks or promote the adoption of longtermist thinking. While many grants so far have prioritized projects addressing risks posed by artificial intelligence (and the grantmakers expect to continue this at least in the short term), the Fund is open to funding, and welcomes applications from, a broader range of activities related to the long-term future.
Thanks!
I agree with you that that’s pretty clear. Perhaps you could just have another sentence explaining that most grants historically have been AI-related because that’s where you receive most of your applications?
On another note, I can’t help but feel that “Global Catastrophic Risk Fund” would be a better name than “Long-term Future Fund”. This is because there are other ways to improve the long-term trajectory of civilisation than by mitigating global catastrophic risks. Also, if you were to make this change, it may help distinguish the fund from the long-term investment fund that Founders Pledge may set up.
Some of the LTFF grants (forecasting, long-term institutions, etc.) are broader than GCRs, and my guess is that at least some Fund managers are pretty excited about trajectory changes, so I’d personally think the current name seems more accurate.
Ah OK. The description below does make it sound like it’s only global catastrophic risks.
Perhaps include the word ‘predominantly’ before the word “making”?
The second sentence on that page (i.e. the sentence right after this one) reads:
“Predominantly” would seem redundant with “in addition”, so I’d prefer leaving it as-is.
OK sorry this is just me not doing my homework! That all seems reasonable.
Which of these two sentences, both from the fund page, do you think describes the fund more accurately?
The Long-Term Future Fund aims to positively influence the long-term trajectory of civilization by making grants that address global catastrophic risks, especially potential risks from advanced artificial intelligence and pandemics. (First sentence of fund page.)
Grants so far have prioritized projects addressing risks posed by artificial intelligence, and the grantmakers expect to continue this at least in the short term. (Located 1500 words into fund page.)
I’d say 2 is clearly more accurate, and I think the feedback you’ve received about donors being surprised at how many AI grants were made suggests I’m not alone.
Could you operationalize “more accurately” a bit more? Both sentences match my impression of the fund. The first is more informative as to what our aims are, the second is more informative as to the details of our historical (and immediate future) grant composition.
My sense is that the first will give people an accurate predictive model of the LTFF in a wider range of scenarios. For example, if next round we happen to receive an amazing application for a new biosecurity org, the majority of the round’s funding could go on that. The first sentence would predict this, the second not.
But the second will give most people better predictions in a “business as usual” case, where our applications in future rounds are similar to those of current rounds.
My hunch is that knowing what our aims are is more important for most donors. In particular, many people reading this for the first time will be choosing between the LTFF and one of the other EA Funds, which focus on completely different cause areas. The high-level motivation seems more salient than our current grant composition for this purpose.
Ideally of course we’d communicate both. I’ll think about if we should add some kind of high-level summary of % of grants to different areas under the “Grantmaking and Impact” section which occurs earlier. My main worry is this kind of thing is hard to keep up to date, and as described above could end up misleading donors in the other direction, if our application pool suddenly changes.