In this 2017 talk, the Open Philanthropy Project's Claire Zabel talks about their work to mitigate Global Catastrophic Biological Risks. She also discusses what effective altruists can do to help, as well as differences between biological risks and risks from advanced AI. At the very end you will find an added section on what you can do to help.
The below transcript is lightly edited for readability.
Today I'm going to talk to you about biosecurity as a cause area and how the Open Philanthropy Project is thinking about it. I think that this is a cause area that EAs have known about for a while, but haven't dug into as deeply as some of the other things that have been talked about at this event - like AI and farm animal welfare and global poverty - but I think it's important. I think it's an area where EAs have a chance to make a huge difference, especially EAs with a slightly different set of skills and interests than those required by some of the other cause areas. I would love to see more EAs engaging with biosecurity.
When I say biosecurity, I want to make sure we're clear on the problem that I'm talking about. I'm focusing on what we're calling Global Catastrophic Biological Risks at the Open Philanthropy Project. I'm going to talk to you about how we see that risk and where we see that risk - where we think it might be coming from. I'm going to talk to you about how I think EAs can make a difference in it. Then I want to note that I'm not really focusing too much on the specific work that we've done and that others have done. I thought it would be more interesting for people to get a sense of what this area is like and the strategic landscape as we see it before getting into the details of specific organizations and people, so hopefully that's helpful for everyone.
I also want to note quickly that I think this is an area where a lot less thinking has been done for a much shorter period of time, so to a greater extent everything should be viewed as somewhat preliminary and uncertain. We might be changing our minds in the near future.
The cause for concern when we think about global catastrophic biological risks is something that could threaten the long term flourishing of human civilization, that could impair our ability to have a really long, really big future full of joy and flourishing for many different sentient beings. That's kind of different from what you might think about biological risks that most people talk about, which are often things like Ebola or Zika. Ebola or Zika are unbelievably tragic for the people afflicted by them, but it doesn't seem like the evidence suggests that they have a realistic chance of causing international civilizational collapse and threatening our long-term future.
To take this further, we predict that we would need a really extremely big biological catastrophe to threaten the long-term future. We're really thinking about something that kills or severely impairs a greater proportion of the entire human civilization than what happened in either of the world wars or in the 1918 flu pandemic. That kind of implies that we're thinking about fatalities that could range into the hundreds of millions or even the billions. There's a lot of really amazing work that could go into preventing smaller risks, but that's not really what we've been focusing on so far. It's not what I anticipate us focusing on in the future. Overall, we're currently ranking the prevention of global catastrophic biological risks as a high priority, although I think it's somewhat uncertain. I think it's high priority to figure out more and then we might readjust our beliefs about how much we should prioritize it.
So what are these risks even like? We think the biggest risks are from biological agents that can be easily transmitted that can be released in one area and spread, as opposed to something like Anthrax, which is very terrible in the space that it's released, but it's hard to imagine it really coming to afflict a large proportion human civilization.
Then within the space of infectious diseases, we're thinking about whether the most risky type of attack would be something that happened naturally that just came out of an animal reservoir, or something that was deliberately done by people with the intention of causing this kind of destruction. Or it might be the middle ground of something that might have been accidentally released from a laboratory where people were doing research.
Our best guess right now is that deliberate biological attacks are the biggest risk. Accidental risk somewhere in the middle, and natural risk is low. I want to explain why that is because I think a lot of people would disagree with that. Some of the reasons I'm skeptical of natural risks are that first of all, they've never really happened before. Humans have obviously never been caused to go extinct by a natural risk, otherwise we would not be here talking. It doesn't seem like human civilization has come close to the brink of collapse because of a natural risk, especially in the recent past.
You can argue about some things like the Black Death, which certainly caused very severe effects on civilization in certain areas in the past. But this implies a fairly low base rate. We should think in any given decade, there's a relatively low chance of some disease just emerging that could have such a devastating impact. Similarly, it seems like it rarely happens with nonhuman animals that a pathogen emerges that causes them to go extinct. I know there's one confirmed case in mammals. I don't know of any others. This scarcity of cases also implies that this isn't something that happens very frequently, so in any given decade, we should probably start with a prior that there's a low probability of a catastrophically bad natural pathogen occurring.
Also, we're in a much better situation than we were in the past and than animals are in some ways, because we have advanced biomedical capabilities. We can use these to create vaccines and therapeutics and address a lot of risks from different pathogens that we could face.
Then finally, on kind of a different vein, people have argued that there's some selection pressure against a naturally emerging highly virulent pathogen because when pathogens are highly virulent, often their hosts die quickly and they try to rest before they die and they're not out in society spreading it the way you might spread the cold, if you go to work when you have the cold.
Now, before you become totally convinced about that, I think that there's some good countervailing considerations to consider about humanity, that make it more likely that a natural risk could occur now than in the past. For example, humanity is much more globalized, so it might be the case that in the past there were things that were potentially deadly for human civilization, but humans were so isolated it didn't really spread and it wasn't a huge deal. Now everything could spread pretty much around the globe.
Also, civilization might be more fragile than it used to be. It's hard to know, but it might be the case that we're very interdependent. We really depend on different parts of the world to produce different goods and perhaps a local collapse could have implications for the rest of the globe that we don't yet understand.
Then there's another argument one can usually bring up, which is if you're so worried about accidental or engineered deliberate attacks, there's also not very much evidence of those being a big deal. I would agree with this argument. There haven't been very many deliberate biological weapon attacks in recent times. There's not a strong precedent. Nonetheless, our best guess right now is that natural risks are pretty unlikely to derail human civilization.
When we think in more detail about where catastrophic biological attack risks come from, we can consider the different potential actors. I don't think that we've really come to a really strong view on this. I do want to explain the different potential sources. Some possible sources could be different states. For example, in bio-weapons programs, states could develop pathogens as weapons that have the potential to be destructive. Small groups, such as terrorists or other extremists might be interested in developing these sorts of capabilities. Individuals who have an interest, people working in various sorts of labs: in academia, in the government and on their own. There are DIY biohacker communities that do different sorts of biological experimentation. Those are the different groups that might contribute to catastrophic biological risk.
There are different kinds of pathogens and I think here - our thinking is even more preliminary - we're especially worried about viral pathogens, because there's proven potential for high transmissibilities and lethality among viruses. They can move really fast. They can spread really fast. We have fewer effective countermeasures against them. We don't have very good broad spectrum antivirals that are efficacious and that means that if we had a novel viral pathogen, it's not the case that we have a huge portfolio of tools that we can expect to be really helpful against it.
I've created this small chart that I think can help illustrate how we divide up these risks. On the top there's a dichotomy of whether the pathogen is more natural or more engineered and then on the vertical axis a dichotomy of whether it emerged naturally (or accidentally) or was a deliberate release. The reason I'm flagging these quadrants is because I think there are two different ways to increase the destructiveness of an attack. One is to engineer the pathogen really highly, and the other is to optimize the actual attack type. For example, if you released a pathogen at a major airport, you would expect it to spread more quickly than if you released it in a rural village. Those are two different ways in which you can become more destructive, if you're interested in doing that. Hopefully you're not. My current guess is that there's a lot more optimization space in engineering the actual pathogen than in the release type. There seems to be a bigger range, but we're not super confident about that.
Here's where the risk we see is coming from. There are advances in gene editing technology, which is a really major source of risk. I think that they've created a lot more room to tinker with biology in general to both lower resources and lower levels of knowledge required and to a greater overall degree, create novel pathogens that are different from what exists in nature, that you can understand how they work. This has amazing potential to do a lot of good but it also has potential to be misused. It's becoming a lot cheaper to synthesize DNA and RNA, to get genetic material for different pathogens. This means that these capabilities are becoming more widely available, just because they're cheaper. Regulating them and verifying buyers is becoming a bigger proportion of the costs, which means companies are more and more incentivized to stop regulating sales and verifying buyers.
Biotech capabilities are becoming more available around the world. They're spreading to different areas, to new labs. Again, this is mostly a sign of progress. People are having access to technology, places in Asia and around the world are having large groups of very talented scientists and that's really great for the most part, but it means there are more potential sources of risk than there were in the past.
Then finally, all of those things are happening much faster than governments can possibly hope to keep up and than norms can evolve, so that leads you to the situation where the technology has outpaced our society and our means of dealing with risk, and that increases the level of danger.
Now I'll contrast and compare biosecurity with AI alignment, because I think AI alignment is something people are much more familiar with. It might be helpful to draw attention to the differences, for getting people up to speed. I think that overall, there's a smaller risk of a far future negative trajectory change from biosecurity. Overall it seems like a smaller risk to me.
With addressing biosecurity risk, there are fewer potential upsides. With an AI, you can imagine that if it develops really well, it has this amazing potential to increase human capabilities and cause human flourishing. With biosecurity, we're basically hoping that just nothing happens. The best outcome is just nothing. No attacks occur. No one dies. Society progresses. In the case of AI alignment, maybe somebody develops an aligned AI, which would be great. But for biosecurity, we're really about preventing downside risks. More of the risk here comes from people with actively bad intentions as opposed to people with good intentions or people who are just interested in the research, especially if you believe and you agree with me that deliberate attacks are the most likely source of concern.
In biosecurity more than AI, I think there are many more relevant actors on both sides, as opposed to there being a few labs with a lot of capabilities in AI. It could be the case that we end up with a situation in biosecurity where there are millions of people that are capable of doing something that would be pretty destructive. And also, we can unilaterally develop counter measures against their attacks. There's less connection between the sources of the risk and the sources of the risk reduction. They're more divorced from one another. There's more possible actors on the sides of attack and defense.
I think that the way that The Open Philanthropy Project is seeing this field right now is somewhat different from how most people are seeing it. Most of the discussion in the field of biosecurity is focused on much smaller risks than the ones that we're worried about. I think discussion of things with greater than one million fatalities was kind of taboo up until very recently. It's been difficult for us to find people that are interested in working on that kind of thing. I think that part of the reason for that, is that it's been really hard to get funding in the space, so people want to make sure their work seems really relevant. And since small attacks and small outbreaks are more common, a good way to make your work more relevant is to focus on those.
There's ongoing debate in the field about whether natural, deliberate or accidental releases are the biggest risks. I don't think people are synced up on what the answer to that question is. I don't think everyone agrees with us that deliberate is mostly the thing to worry about. Then people are really trying to walk this tightrope of regulating risky research while not regulating productive research, maintaining national competitiveness, and encouraging productive biotech R&D.
Given all of that, we have some goals in this space. They're kind of early goals. They won't be sufficient on their own. They're mostly examples, but I think they could get us pretty far. The first thing is we just really need to understand the relevant risks in particular. I'm keeping it very high level for now, because there's not a lot of time, and partly because I think that talking about some of these risks publicly is not a productive thing to do, and also because we're pretty uncertain about them. I think it would be really helpful to have some people dig into the individual risks. Think about what one would need to do in order to pull off a really catastrophic bio attack. How far out is that being a possibility? What sorts of technological advancements would need to occur? What sorts of resources would one need to be able to access in order to do that? If we can answer these questions, we can have a sense of how big catastrophic biosecurity risks are and how many actors we need to be worried about.
Understanding particular risks will help us prioritize things we can do to develop counter measures. We want to support people in organizations that increase the field's ability to respond to global catastrophic biological risks. The reason for that is that right now the field of biosecurity has lacked funding for a long time. A lot of people have left the field. Young people are having a very difficult time going into the field. Hopefully that's changing, but it's still a pretty dire situation, in my view. We want to make sure that the field ends up high quality with lots of researchers that care about the same risks we care about, so people that show signs of maybe moving in that direction, we're very enthusiastic about supporting, in general.
Then finally, we want to develop medical counter measures for the things that we're worried about. We've started having our science advisors look into this. We have some ideas about what the worst risks are and if we can develop counter measures in advance and stockpile those, I think we would be much better prepared to address risks when they come up.
Finally, I want to talk to you a little bit about what I think EAs can do to help. I see a lot of potential value in bringing parts of the EA perspective to the field. Right now there aren't a lot of EAs in biosecurity and I think that the EA perspective is kind of special and has something special to offer people. I think some of the really great things about it are, first of all, the familiarity with the idea of astronomical waste and the value of the far future. That seems like it's somewhat hard to understand. It's a bit weird and counterintuitive and philosophical, but a lot of EAs find it compelling. A lot of other people find it wacky or haven't really heard about it. I think having more concern about that pool of value and those people in the future who can't really speak for themselves could do the field of biosecurity a lot of good.
Another thing that I think is amazing about the EA perspective, is comfort with explicit prioritization, the ability to say, "We really need to do X, Y, and Z. A, B, and C are lower priority. They'll help us less. They're less tractable. They're more crowded. We should start with these other things." I think right now, the field doesn't have a clear view about that. There's not a very well thought out and developed road map to addressing these concerns. I think EAs would be good at helping with that.
Finally, I think a lot of EAs have a skepticism with established methods and expertise. That's great because I think that's necessary actually in almost every field. Especially in fields that involve a complicated interplay of natural science and social science. I think that there's a lot of room for things to be skewed in certain directions. I haven't seen too much harmful skew, but guarding against it would be really helpful.
There's some work going on at the Future of Humanity Institute that we're very excited about. It seems like there's a lot of low hanging fruit right now. There are a lot of projects that I think an EA could take on and they'd be pretty likely to make progress. I think biosecurity progress is more of a matter of pulling information together and analyzing it, and less based only in pure insight.
I think that you should consider going into biosecurity if you are an EA concerned with the far future, who wants to make sure that we all get to enjoy our amazing cosmic endowment, and if you think that you might be a good fit for work in policy or in the biomedical sciences.
This is an area where I think that a lot of safety might come from people not overhyping certain sorts of possibilities as they emerge, at least until we develop counter measures. It's important to have people that feel comfortable and are okay with the idea of doing a lot of work and then not sharing it very widely and actually not making it totally open, because that could actually be counterproductive and increase risk. That's what I hope that people will be willing to do. I hope that we find some EAs who want to move into this field. If you feel like you're interested in moving into this field, I would encourage you to reach out to me or grab me sometime at this conference and talk about both what you'd like to do and what might be stopping you from doing it.
In the future we might write more about how we think people can get into this field and be able to do helpful research, but we haven't really done that yet, so in the meantime, I really hope that people reach out. Thank you so much and I'll take your questions.
Question: Okay, so we've got a number of questions that have come in and I'm just gonna try to rifle through them and give you a chance to answer as many as we can. You emphasized the risk of viral pathogens. What about the, I think, more well known if not well understood problem of antibiotic resistance? Is that something that you're thinking about and how big of a concern is that for you?
Claire Zabel: Yeah. I think that's a good question. The Open Philanthropy Project has a report on antibiotic resistance that I encourage you to read if you're curious about this topic. I think it's a really big concern for dealing with conventional bacterial pathogens. Our best guess is that it's not such a special concern for thinking about global catastrophic biological risks, first of all, because there's already immense selection pressure on bacteria to evolve some resistance to antibiotics, and while this mostly has really negative implications, it has one positive implication, which is that, if there's an easy way to do it, it's likely that it'll happen naturally first and not through a surprise attack by a deliberate bad actor.
Then another reason that we're worried about viruses to a greater extent than bacteria is because of their higher transmissibility and the greater difficulty we have disinfecting things from viral pathogens. So, I don't think that antibiotic resistance will be a big priority from the far-future biosecurity perspective. I think it's possible that we're completely wrong about this. I'm very open to that possibility, and what I'm saying is pretty low confidence right now.
Question: Great. Next question. To what extent do small and large scale bio-risks look the same and to what extent do the counter measures for those small and large scale risks look the same, such that you can collaborate with people who have been more in the traditional focus area of the smaller scale risks?
Claire Zabel: That's an interesting question. I think it's a complicated one and a simple answer won't answer it very well. When I think about the large scale risks, they look pretty different for the most part from conventional risks, mostly because they're highly engineered. They're optimized for destructiveness. They're not natural. They're not something we're very familiar with, so that makes them unlikely to be things that we have prepared responses to. They're likely to be singularly able to overwhelm healthcare systems, even in developed countries, which is not something that we have much experience with.
But the second part of the question about the degree to which efforts to address small scale risks help with big scale risks and vice versa, I think that that's somewhat of an open question for us and as we move towards prioritizing in the space, we'll have a better view. There's some actions that we can take. For example, advocacy to get the government to take biosecurity more seriously might help equally with both. On the other hand, I think developing specific counter measures, if we move forward with that, will be more likely to only help with large scale risks and be less useful with small scale risks, although there are counter examples that I'm thinking of right now, so that's definitely not an accurate blanket statement.
Question: When you think about these sort of engineered attacks that could create the largest scale risk, it seems like one thing that has sort of been on the side of good, at least for now, is that it does take quite a bit of capital to spin up a lab and do this kind of bioengineering. But, as you mentioned, stuff is becoming cheaper. It's becoming more widely available. How do you see that curve evolving over time? Right now, how much capital do you think it takes to put a lab in place and start to do this kind of bad work if you wanted to and how does that look five, ten, twenty years out?
Claire Zabel: I don't think I want to say how much it takes right now, or exactly what I think it will take in the future. I think the costs are falling pretty quickly. It depends on what ends up being necessary, so for example, the cost of DNA synthesis is falling really rapidly. It might be the case that that part is extremely cheap, but actually experimenting with a certain pathogen that you think might have destructive capability - for example, testing it on animals - might remain very expensive, and it doesn't seem like the costs of that part of a potential destructive attack are falling nearly as quickly.
Overall, I think costs will continue to fall but I would guess that the falling plateaus sometime in the next few decades.
Question: Interesting. Does biological enhancement fall within your project at all? Have you spent time considering, for example, enhancing humans or working on gene editing on humans and how that might be either beneficial or potentially destabilizing in its own way?
Claire Zabel: That's not something that we've really considered a part of our biosecurity program.
Question: Fair enough. How interested is Open Philanthropy Project in funding junior researchers in biosecurity or biodefense? And relatedly, which would you say is more valuable right now? Are you looking more for people who have kind of a high level strategic capability or those who are more in the weeds, as it were, of wet synthetic biology?
Claire Zabel: Yeah. I think that right now we'd be excited about EAs that are interested in either, potentially, depending on their goals in this field, the extent of the value alignment, and their dedication and particular talents. I think both are useful. I expect that the kind of specialization, for example, either in policy or in biomedical science will possibly be more helpful in the long term. I'm hoping that we'll gain a lot of ground on the strategic high level aspects of it in the next few years, but right now I think both are sorely needed.
Question: Next question. For someone whose education and skills have been focused on machine learning, how readily can such a person contribute to the type of work that you're doing and what would that look like if they wanted to get involved?
Claire Zabel: I don't know. I've never seen anyone try. I think that it would be possible because I think that there's a lot of possibility of someone who has no special background in this area, in general, becoming really productive and helpful within a relatively short time scale and I don't see machine learning background as putting anyone at a particular disadvantage. Probably it would put you at somewhat of an advantage, although I'm not sure how. I think that right now, the best way to go would probably be just to get a Masters or PhD in a related field and then try to move into one of the relevant organizations, or try to directly work at one of the relevant organizations like our biggest grantee in biosecurity, the Center for Health Security. And for that, I think that probably having a background in machine learning would be neither a strong drawback nor a huge benefit.
Question: That's about all the time that we have for now, unfortunately.