← 80,000 Hours Podcast
Loading summary
Transcript247 lines
- [00:00]
A
Myrobacteria might be able to persist in the environment directly, so they might be able to grow in soil or in oceans. It could be like living on Earth today without an immune system, or even like living on Earth today with an immune system, but where you could catch Ebola from trees or from your pet cat to cause harm, all a mirror bacterium needs to be able to do is grow. How much would it cost to make mirror life from where we are today? People estimate some something like 500 million to a billion dollars would be sufficient. That motivates the need to have these discussions now. I think at the moment the marginal impact of an extra person working on mirror life is huge. You could probably become the expert in policy around mirror life in your country within a few weeks or months.
- [00:49]
B
Today I'm speaking with James Smith. James is the co director of the Mirror Biology Dialogues Fund and an adjunct associate professor at J. Craig Vent Institute. James was also one of the authors of the kind of bombshell science paper that drew attention to the risks of mirror life last December. Thanks for coming on the podcast, James.
- [01:10]
A
It's great to be here. Thanks for having me.
- [01:13]
B
Okay, so in the science paper you and your co authors basically make the argument that we could, we're like reasonably close to creating what's called mirror life. And mirror life is, is basically like normal life, but the molecules are arranged backwards kind of in the mirror image. And despite this seeming like actually quite a trivial difference, it makes these life forms, which could be bacteria, extremely, extremely dangerous to not just humans, but like most species, including animals and plants. What exactly makes them so dangerous?
- [01:59]
A
Yeah, it might seem like a trivial difference, but actually this is a really fundamental break from normal biology. So ever since the last Universal Common Ancestor 4 billion years ago of all life on Earth, all of our DNA has always been made up of right handed building blocks, meaning that it makes double helix that twists to the right. And mirror bacteria would have the opposite. They'd have DNA that twists to the left. So introducing it to Earth would be a bit like introducing an alien invasive species that nothing on Earth has evolved to deal with. To take one specific example, our immune systems need to be able to recognize pathogens when they get into the blood to be able to mount an effective response. The way that it does that is through receptors that you could think of like gloves. Now putting a right hand in into a right handed glove, it fits well, it activates the immune system. But if you try and put a left hand into a right handed glove, it doesn't work properly. Mirror life would be a bit like that. A lot of the interactions that are so fundamental to biology are not going to work in the same way.
- [03:06]
B
Yeah, I mean, it's wild. It just doesn't seem intuitively like this should matter. It's the same molecule or the same life form. But because there are all of these mechanisms that most of us never think about that actually are like, yeah, the glove has to fit inside the. The hand has to fit inside the glove. Actually, really, really important. Is there more we should talk about in terms of what mirror life actually is?
- [03:39]
A
Yeah. Mirror life would be a type of artificial life that you'd make in the lab. A mirror bacterium is what we're mostly going to talk about today, because that would be the simplest form of life that you'd make. And it would literally have exactly the same components as a normal bacterium. So it would have DNA, proteins, ribosomes, a membrane, but all of those would be made in their mirror image form.
- [04:04]
B
And you think that this problem is kind of on par with some of the most pressing global problems we know of, like AI safety and also kind of other biosecurity threats. Why is this comparable? Especially I think a thing that's surprising is it feels very new. This paper was quite recent. It feels quite surprising that we've discovered a new kind of science that happens to be one of the most threatening things to humanity and the earth.
- [04:39]
A
Yeah. People have actually posited the existence of mirror life since the 1800s. So it's not a completely new thing. People have mentioned the risks of it in kind of passing in the literature, but it wasn't until last year that anyone looked into this properly. And I think this is the worst bio threat that I've seen described. And I think it can be on par with AI safety for some people, depending on your skill set. That's driven a lot by the fact that there are literally probably 10 people working full time to think about the risks for myrobacteria and what could be done to address them. So it's incredibly neglected. I also think it's really tractable. We're used to thinking about technologies like artificial intelligence or nuclear, where you have massive risks but also potentially massive benefits. And Mirror Life seems to be an exception to that. The benefits that we can foresee are really quite minor, whereas we have these massive risks. And I don't think there are going to be strong commercial drivers to want to make mirror Life. And that means I think we have a really good chance of at kind of cutting off this risk.
- [05:40]
B
Okay, so to make this a bit more concrete and kind of drive home the importance of this issue, you've been trying to raise the alarm about it, but imagine that you fail and kind of your worries about what would happen if mirror life were created and released into the environment kind of came true. How would that play out? What would it look like?
- [06:02]
A
Obviously, this is skipping out a lot of nuance, but. But it could be like living on Earth today without an immune system, or even like living on Earth today with an immune system, but where you could catch Ebola from trees or from your pet cat or from a carrot that you eat or something like that. So it could be really crazy. More concretely, I'd break it down into two areas of risk. One is immunological, and the second is ecological. Okay, so Myobacteria could infect not just humans, but. But a wide range of different species because they have this property of broad immune evasion. We're used to thinking about something like Covid or influenza that can spread from human to human, but it doesn't also infect plants and insects, livestock, whereas myrabacteria might be able to do that. And the second area is around the ecological risks. Myrobacteria could grow and might be able to persist in the environment directly. So they might be able to grow in soil or in oceans, and that opens up another transmission route directly from the environment to multicellular hosts. So you might be able to get infected by mirror bacteria from dust blowing into your home that has bacteria on it that you inhale or something like that. And we don't really have the tools and technologies to deal with a threat like that at the moment.
- [07:22]
B
Yeah. Just to make sure I understand when you say living without an immune system, that is really, really importantly different to kind of living with a normal immune system, but encountering a new pathogen that we haven't seen before. So Covid, I guess, is a kind of example where we're exposed to new kind of versions of pathogens all the time, but our immune system is able to figure out something, some kind of response that allows us to, for the most part, defeat those kind of infections. In the case of mirror life, is there an analogy to understand how exactly it's different from that? Like, it's not that it's living without any immune system at all?
- [08:15]
A
Yeah. I mean, there might still be parts of the immune system that work, but there are examples of people that have had immune disorders that essentially replicate what it would be like to not have an immune system. There was, for example, A guy in the 70s called David Vetter, who was also known as Bubble Boy. When he was born, he had to be immediately put into a sterile plastic chamber where he lived all 12 years of his life. All of his food, any of the toys that he used, anything like that before entering that plastic chamber, needed to be sterilized. So there's a proof of concept that we could live in a world like that, but I don't think it's the world that we should be aiming for.
- [08:51]
B
Yeah, I mean, that is horrific. And the thing there is like, yeah, it's not that we have to kind of learn the specific immune response that's going to work for this particular pathogen. It's like, most likely we are fundamentally not going to be able to. And so this, in this case, this person, Bubble Boy, lived in this environment for the duration of his life because his immune system was never going to figure out how to like tolerate bacteria and respond to them effectively. And that is what mirror life would be like, because it would be so our immune system would be so kind of just like completely ill equipped to dealing with it.
- [09:38]
A
Yeah, that sounds about right. There's a lot of nuance here that we're going to unpack. We will get into it that we're going to unpack for sure. But I think this is a reasonable analogy, and we know from a lot of people who have defects in just one major pathway of immunity that they're similarly really susceptible to bacterial infections. So even if one or even many parts of the immune system do work, that's not enough to be confident that you're going to be protected from an infection.
- [10:05]
B
Yeah, okay, I want to get into that. But first, why would anyone make mirror bacteria given that there's massive downside? And we'll talk a bit about the nuance about this, but not much upside?
- [10:17]
A
Yeah, I mean, historically, I think the main reason people wanted to do this was because it would be a really cool technical project. And I think there is something aesthetically very pleasing about creating this mirror of, of life on Earth that I completely buy into. But I don't think it's sufficient justification when there are these risks. People didn't know about these risks until quite recently. So now I don't think any well meaning scientist is going to go ahead and build mirror life. And in fact, all the people who previously said they wanted to make mirror life have been part of this discussion and are calling for mirror life to not be made. So what I'm most concerned about now is, is malicious actors that Might in the long term, want to make this to cause a huge amount of harm. Also, the possibility that AI could do this, this is something that's been recognized in a fair amount of AI commentary now, like AI 2027, they talk about neurobacteria in the slowdown scenario. Will MacAskill talks about this in his Preparing for an Intelligence Explosion. And a bunch of other senior AI scientists have been referencing this work. So that's something that I think needs more thought. It's also possible that someone might develop mirror bacteria for industrial applications, but at this stage, I think that's quite unlikely.
- [11:33]
B
Okay, how close are we? My understanding is that we can't make mirror bacteria yet, but we're making progress toward it. Does that mean it's like a year away or is it decades away?
- [11:49]
A
If you ask the synthetic biologists, most of them who work on this, they'll say something like 10 to 30 years away. But I don't think that's really pricing in the potential for transformative AI to speed things up significantly. Another way to think about this is how much would it cost to make mirror life from where we are today? And there people estimate something like 500 million to a billion dollars would be sufficient, and that could make it happen much more quickly. And there are people like George Church who think this might happen literally in the next couple of years. So there's a lot of diversity here. I think it's really hard to be confident in any particular estimate. And I think that motivates the need to have these discussions now about what we should do to address the risk.
- [12:31]
B
How did you end up involved in this work? My sense is that you were already doing kind of work to reduce biological risks in the world, but not involved in this work. Something like over a year ago.
- [12:46]
A
Yeah, I got involved in this middle of last year. Before that, I was working on AI enabled biology, doing some little bits on AI safety and actually thinking that I would work on AI safety in the longer term. But within two weeks of hearing about this issue, I quit the other work I was doing to focus full time on this and have been doing that ever since.
- [13:08]
B
Wow.
- [13:09]
A
And the main reasons for that were the tractability and neglectedness in combination with the scale of this problem. So when I first got involved in the project, I worked with John Glass and Jack Szostak, two leading really top scientists, to coordinate this working group of scientists and write the paper that brought this risk to the attention of the world.
- [13:32]
B
Okay, and did you? My sense is that this is especially true among the scientists who really wanted to make mirror life because it was super cool. But that in general it's been true that when scientists first hear about this, there's at least some skepticism that mirror life will be as catastrophic as you've kind of just laid out. Did you have any of that skepticism and was there anything in particular that convinced you?
- [14:03]
A
Yeah, I was definitely sceptical initially. I think pretty much everyone who first hears about this is. And I think that's understandable. Like base rate, some crazy science thing. People say it's risky, it's going to end the world. It's probably not.
- [14:18]
B
Yeah, yeah.
- [14:19]
A
But the difference here is the amount of work that went into really thinking through this in detail. So I think two things really convinced me. One. One was the fact that some of the best scientists in the world, some of the best immunologists, ecologists, synthetic biologists, but also biosecurity experts and others had been looking into this specifically being asked, what's the problem with the analysis? Where is there a hole in this? How can we rule out this catastrophic risk? And they hadn't been able to. And that I found pretty compelling. And then the other thing was just imagining what it would actually be like to live in a world where myrobacteria existed. If this analysis was right, any exposure to the outside world could end up being fatal. You could catch it from plants, whole ecosystems could be destroyed. That kind of blew my mind. And I thought, okay, this is something I need to get involved in.
- [15:13]
B
Yeah. I think a thing that really struck me was like, these are some of the most accomplished scientists of our time. It's people who have made extremely groundbreaking discoveries. It's not like, I don't know, random scientists that we haven't heard of and they don't work on anything. Super mind blowing.
- [15:35]
A
Yeah, it was pretty wild and is pretty wild to get to work with all these amazing people.
- [15:40]
B
Let's talk more about what makes mirror bacteria and I guess mirror life more broadly, potentially so catastrophic. So when a normal bacteria enters the body, it wants kind of what any organism wants. It wants to survive and reproduce. So eat and kind of double itself continuously. And in kind of normal human bodies, for most people, the immune system responds by kind of mounting a counterattack with white blood cells and the whole immunity thing. What happens when mirror life enter the body?
- [16:21]
A
We're used to thinking about the immune system as being able to deal with kind of any arbitrary threat that's thrown at it. And that's kind of the case for most things that actually do get thrown at it. But the difference with mirror life is that we haven't evolved to be able to deal with mirror life because so much of the immune system depends on specific binding between molecules that need to have a certain shape. And with mirror life, the molecules that matter are going to be reversed, and so a lot of this binding isn't going to work. So to take an example, in the innate immune system, which is kind of the first line of defense that the immune system has for that to be activated, you have these things called pattern recognition receptors, which basically bind to molecules on invading bacteria. And those molecules on the invading bacteria are chiral. So examples would be bacterial DNA or flagellin, which is a protein in the bacterial tail. And in mirror bacteria, those would be reversed. So they wouldn't bind properly to the pattern recognition receptors. And that means that the innate immune system wouldn't be activated properly. That's important because the immune system is actually highly interdependent. Ruslan Medsov, who was one of the co authors on the science paper, discovered that the adaptive immune system which mounts your antibody response depends on the innate immune system to work. So just breaking this one part, it can kind of break the whole immune system. But with mirror bacteria, it wouldn't just be breaking this one part. There are other parts that we think wouldn't work too. And we can look at, there are lots of analogies from humans and mice that we can look at to sort of see how that plays out in experiments.
- [18:18]
B
Cool. Okay. Yeah. Just to make sure I understand, the innate immune system kind of recognizes foreign things like bacteria by having, I don't know, I guess a lock and key. Feels like a very common analogy. So it's got a lock with a certain shape and a mirror. A normal bacteria has a tail that fits in that shape. And then when they come together, the immune system is like, it's you, I recognize you. I'm going to mount an attack because I know that you give me illness. But the mirror bacteria has kind of the mirror image tail and it just doesn't fit.
- [19:01]
A
I think what you described is the way that chirality is important in general across a lot of biological interactions. I think the best analogy is probably a hand in a glove. So if you think about an immune receptor as a glove, let's say it's a right handed glove. Right handed gloves fit right hands very well. And the pathogen's molecules, let's say they're the right hands. If instead you're trying to put a left hand into a right handed glove, it's not going to go in properly. I mean, maybe it will go in a bit, which might actually be the case with some of this, but it's not going to fit as well. And that's kind of the fundamental issue. The immune system isn't binary. So just one part of the immune system working is not necessarily going to be enough. And we know from human immune disorders that if one important part of the immune system breaks, then people will die in childhood. So for MHC class II deficiency, patients that have this disease usually die before the age of 10. About 40% of them die before the age of 10 unless they can get a curative bone marrow transplant.
- [20:05]
B
Wow.
- [20:06]
A
And that's just one part of the immune system not working with a mirror. Bacteria infection, that part is very unlikely to work, I think. But also the adaptive immune system, the pattern recognition receptor binding, is unlikely to work. And again, there are disorders that mimic that. So there's a disorder called MYD88 deficiency, which basically means that the pattern recognition receptor repertoire doesn't function properly. And in patients that have that, something like 30% of them die before the age of two.
- [20:36]
B
Oh, God. Okay.
- [20:38]
A
It's quite bleeding.
- [20:40]
B
Keep in mind, how do I guess these kind of diseases of the immune system and mere bacteria, in theory, cause humans to die?
- [20:52]
A
Well, we don't know for sure. It's pretty hard to know because we haven't ever seen anything that's exactly like this. But one way to think about it is kind of like weeds taking over a garden. They're not deliberately causing harm, but they're just growing and spreading and using up nutrients. And eventually, if something is growing and spreading in your body, it's going to cause an issue. One way that that might play out is with something that looks like sepsis. Sepsis is a common cause of death. If you have a bacterial infection, it's actually the third most common cause of death in the US Once you're in the hospital. So it's pretty common. So it's pretty common, but it's also not that well understood. But basically, if you have my bacteria in your blood, they're going to be replicating using up nutrients, and eventually those nutrients, they would have been used by your host cells for something. There's a reason that they're there.
- [21:49]
B
Yeah.
- [21:50]
A
That's going to eventually start to cause some of your host cells, let's say in one of your organs, like your liver or something, to start to die. And your immune system, your immune system's pretty amazing. Like it can detect foreign things. But it can also detect when your own cells, when their insides are outside of them. So that would start to happen. These cells are dying, and they're releasing their insides out into the body, and that causes an immune response. But when people have bacterial infections that are quite severe, that immune response often gets out of control and starts a positive feedback cycle where the immune system overreacts, and that ultimately causes people to die.
- [22:36]
B
Wow. The irony of that. I know, like, the immune system is like, I'm going to help, and then actually is extremely unhelpful.
- [22:48]
A
Yeah, it's pretty unfortunate. Another way that it might happen is kind of if they got to extremely high concentrations, which would probably not usually happen with something other than myrobacteria, it might physically start to block things or result in there not being enough oxygen in the blood for you to survive. So there could be some really unusual pathologies.
- [23:11]
B
I guess a thing that I find kind of strange about this is it just feels really counterintuitive that mirror bacteria could have this catastrophic effect on the human body. But the human body is kind of powerless and defenseless against mirror bacteria. Why is there this asymmetry? Is there some intuitive way of understanding that?
- [23:35]
A
I think it is quite counterintuitive, but it just kind of happens that when you get into the details of what's going on, it seems to be true. I think one way to think about it is that to cause harm, all a mirror bacterium needs to be able to do is grow. And we can be pretty confident that it would be able to grow because we know that there are enough achiral nutrients in the body. So ones that don't come into mirror image forms, and as long as it can grow, it's ultimately going to cause harm. Whereas for a human immune system to be able to respond to a mirror bacterium effectively, it needs to be able to recognize that it's there, and the processes by which it recognizes the myrobacterium are likely to fail.
- [24:21]
B
I guess if you think about kind of what a mirror bacteria's goals would be, eating, reproducing, reproducing, it seems like there'd be no issues there. But maybe I can imagine why a mirror bacteria might, like, maybe it needs to consume chiral, like, mirror image nutrients. To what extent would that be a barrier to mirror bacteria infections getting really out of control?
- [24:54]
A
Yeah. The first question I had when I heard about mirror life was, I don't get it. What would it eat?
- [25:00]
B
Okay.
- [25:01]
A
I think this is a question that a lot of people have, actually. So a lot of nutrients do come in left and right handed forms, but not all of them. Some of them are achiral. And an example of an achiral object is a sphere. So you reflect it in a mirror and it looks exactly the same. And if you think about the hand into a glove analogy, this would be more like a hand holding a ball. The ball kind of looks the same regardless of which hand you are. And nutrients can be like that too. So in the human blood, you have things like acetate, acetoacetate, glycerol, succinate, pyruvate. All of these are achiral nutrients that are present in quantities that it looks like would be enough to support growth. And it turns out that there's really interesting experimental evidence from E. Coli, natural chirality E. Coli. So just normal E. Coli, a type of bacterium that's very commonly studied that you can grow it on completely achiral carbon sources as food. So you can basically feed it only achiral food sources, and it will still grow.
- [26:04]
B
I see, okay.
- [26:05]
A
And from the perspective of a mirror bacterium, an achiral food source looks exactly the same as it does from the perspective of a normal chirality E. Coli. So you can just directly infer that the myrrhobacterium would be able to grow on the same things. And then we know that those nutrients, those achiral nutrients that E. Coli have been shown to be able to grow on, are present in human blood at concentrations that would be sufficient to enable growth. There is some uncertainty here. So they're going to have some fitness disadvantage. They're going to probably grow a bit more slowly than they might otherwise. But this is assuming that we're just taking an exact mirror of a natural E. Coli. The thing I'm most worried about is a malicious actor doing this. In that case, if you're trying to cause harm, you would engineer into that bacterium the ability to consume common chiral nutrients like glucose. And in that case, you kind of. In that case, you would no longer have this nutritional constraint. And I think it's pretty hard to get out of the conclusion that if someone was trying to cause harm, they wouldn't be able to engineer around some of these constraints. There's already a blueprint for how you would introduce the ability to consume D glucose, which is the common form of glucose. There's a alpha proteobacterium that can metabolize mirror glucose. So there's one that we already know about in the world that can metabolize mirror glucose. And so you could Use that pathway, but in its mirror form and engineer that into great. Into your mirror bacterium. And then it would be able to consume. To consume normal glucose. And that's a very, very common nutrient. Something we haven't talked about is if you were to make mirror bacteria for industrial applications, you probably also would want to engineer them so that they could consume common nutrients, because those nutrients are going to be easier to get hold of to grow it. So let's say you were making myobacteria to produce drugs, which is one thing that people have been interested in, then you'd probably want to engineer the bacterium to be able to consume glucose. And that means that the risk of an accident there would be high.
- [28:19]
B
Yep. Makes sense. Bad news all around. Okay, is there an intuitive way to understand, I guess, like innate and adaptive parts of the immune system? Just for the sake of like, I feel like. I feel like an intuitive analogy might help me keep it straight as we talk about this more. Is innate something like over many centuries and much longer, our bodies have evolved the ability to recognize certain bacteria or other organisms that have been around for a long time. And adaptive is something like we notice new things in the environment that we haven't been exposed to before and we figure out a way to then mount an immune response to those new things, even though we haven't kind of evolved specific lock and key mechanisms for those particular pathogens. I'm basically guessing based on the words innate and adaptive.
- [29:23]
A
Yeah, that's pretty much right. So the innate immune system, one way to think about it, is the first line of defence. And it works by recognizing patterns that are common to pathogens. So, for example, it recognizes bacterial DNA. All bacteria have bacterial DNA. It recognizes lipopolysaccharide, which is really common on the surface of bacteria. Adaptive immunity is kind of a specialized defense. In theory, it can mount a response to any arbitrary invader. So you have antibodies that have so much diversity that they can basically bind to any potential surface, including actually to mirror proteins. Some of the key cells in adaptive immunity to be aware of are T cells and B cells.
- [30:12]
B
Let's turn to animals and plants and I guess, kind of ecosystems more broadly. How would mirror life do kind of when exposed or when an animal, a non human animal, is exposed to it.
- [30:28]
A
So for most vertebrates have immune systems that are quite similar to human immune systems. So the last common ancestor of the vast majority of vertebrates that are alive today had an innate immune system that has the characteristics that we talked about with pattern recognition receptors that play a key role and had adaptive immunity based on T and B cells. So that means that the deficiencies that we're expecting in humans are likely to be applicable to a lot of other animals as well. But there is diversity in animals, so I don't think this is universally going to be the case. Some of them will have different susceptibility. The Atlantic cod, for example. The Atlantic cod has lost the ability to present antigens via MHC class II and instead has an expanded set of these pattern recognition receptors. And the zebrafish, which is a really commonly studied model organism for developmental biology, has like 10 times the number of NOD like receptors, which are a type of pattern recognition receptor, than humans and mice. And most of these receptors are not characterized, so we don't know what they bind to. They might bind to achiral things. And more generally, I think it's very likely that some animal immune systems will just happen to work against this. Having said that, I still think it's plausible that a large fraction of animals could be susceptible here. There are even some vertebrates like the lampreys and hagfish, which, if you've never seen a picture of them, they look really like the sandworm from Dune. They have slightly different immune systems which we haven't looked into in as much detail, but share the same principles. Right, that's vertebrates, then invertebrates. So things like insects. The immune systems are less well characterized. So insect immune systems are less well understood than vertebrate ones. But we do know that insects have innate immunity. That's quite similar to what we have in vertebrates. They lack an adaptive immune system, but they do have innate immune systems that similarly rely on pattern recognition receptors to get them to work. So in fruit flies that are one of the most well studied examples, their antibacterial defenses being activated is downstream of binding of a pattern recognition receptor to a molecule called peptidoglycan, which is a chiral molecule that's present in bacterial cell walls. And there have been experiments where these fruit flies have that receptor knocked out, so it doesn't work. And that really increases the susceptibility of the fruit flies to infection by bacteria. So common bacteria that don't normally cause disease will end up killing them. And this is also true in mosquitoes and bees. Beyond that, I'm not so sure. There's a lot of this that hasn't been studied in detail. But the same principles of immunity are common to all animals. And so I think we should unfortunately be quite worried.
- [33:48]
B
Yeah, grim.
- [33:49]
A
I'm somewhat confident that there'll be at least something that would be able to respond to this. But in the animals that we've studied, the best humans, we, I think, do know enough to be quite worried. Even in humans, there's definitely a possibility that we're missing something in the analysis. And in fact, the immune system can deal with it from some unpredictable way. It's going to be very difficult to rule that out. But many of the world's best immunologists have looked into this and have been unable to rule out this scenario where it causes harm to humans. These are people like Ruslan Medzitov, who discovered the fact that the adaptive immune system depends on activation of the innate immune system. Mark Davis, who discovered T cell receptors. David Relman. Yeah, David Relman, who was the first person to characterize the human microbiome. Just like huge names in the field have looked into this and decided to co author this paper, calling attention to the risks. And they haven't been able to find a knockdown argument for why we shouldn't be concerned.
- [34:57]
B
Okay, and how about plants? I guess I can imagine there being a similar story, but I can also imagine plant immune systems being very, very different to animal ones.
- [35:08]
A
Yeah, plant immune systems are actually conceptually quite similar to animal ones. They don't have adaptive immunity, but they have innate immunity, similar to insects. And that innate immunity similarly has these pattern recognition receptors that detect common patterns on bacteria. So to give a specific example, plant leaves have little holes in them called stomata. The leaf surface has these pattern recognition receptors that will bind to common patterns on bacteria. When they do that, the stomata will close up in something called stomatal defense to stop the bacteria from being able to get into the leaf.
- [35:47]
B
Right, right.
- [35:48]
A
That probably wouldn't work as well with a myrobacterial infection. And so the bacteria might be able to enter the leaf through the stomata. Once it's inside the leaf, you have this space in between cells called the apoplast. And the cells kind of around that space, again, have these pattern recognition receptors on their surface that let them detect bacteria getting into the leaf. If they detect them, then they mount a response, releasing things like reactive oxygen species. And so, again, that might not work as well. But a key difference between plants and animals is that plants rely a lot more on physical barriers for defense. And Myrobacteria, by default, would lack the specialized enzymes to break down parts of the plant which. Which might be necessary for it to move between leaves or to get into the plant vasculature. So myrobacteria might not be able to get into the xylem or the phloem, which are like the veins of the plant, and then move around it. So it's kind of unclear whether it will end up causing serious harm to plants. I think there's a reasonable chance that it would. But plant immune systems generally are much less well understood. Again, some of the top people have looked into this. Jonathan Jones, who discovered a lot of parts of plant immunity, is one of the co authors on the paper. But I think there is more uncertainty here than in the case of animals.
- [37:12]
B
Okay. And so the thing that seems promising for plants is whereas animals have more of this lock and key function, plants have more of like brute forcing, like brick walls and mirror bacteria are. There's like the fact that they're this other shape doesn't make them better able to get around that. They just also might struggle getting around the brick wall.
- [37:42]
A
Yeah, that's right. I mean it's still possible that they would, to be clear. So myrobacteria could end up being able to kill lots of plants, including crops. And I think we are going to struggle to rule that out. One way that this might happen is common way that bacteria will get into the phloem, which is one of the veins in plants, is through phloem feeding insects. So the insects kind of have the bacteria in their salivary glands and then they're feeding on the phloem, which will have SAP in it, which is kind of the blood equivalent in plants. And then the bacteria get from the insect into the phloem. That could happen if you're having insects that have been infected with Myobacteria.
- [38:24]
B
Okay, yeah. God, I can see how there's so much uncertainty because there are some reasons to be optimistic in all of these different cases, maybe in some cases more than others, but would not have thought of insects that are infected because they do have similar immune systems to other vertebrates could then get around the brick wall. And that's. And that's how plants are infected. So yeah, it's a bit of a minefield.
- [38:57]
A
Isn't was kind of surprising to me initially how common these features of immunity are across all different multicellular life. But I think the thing is that no multicellular life has had a reason to evolve to be able to deal with mirror life because it's never interacted with it. So in a way it makes sense that this would be an evolutionary blind spot for it. There's no reason why it should be able to detect It. But I think a lot of people initially have an intuition that the immune system is just really good at dealing with any arbitrary threat.
- [39:30]
B
Right.
- [39:31]
A
In fact, it seems like that wouldn't be the case here.
- [39:34]
B
Can you explain the kind of broader environmental risks?
- [39:39]
A
Yeah, definitely. Myrobacteria could really damage a lot of habitats, drive potentially a lot of multicellular species to extinction, and might even change things like geochemical cycling. So we've talked about the. Yeah, we've talked. We'll get into all of it, I hope. But we've talked about the immune system defects. So there's the direct impact on animals that could, of course, impact ecosystems. But Myobacteria would also escape common forms of predation. So normal bacteria are killed by viruses all the time. There's about 10 times the number of bacteriophages, which are viruses, for bacteria, than there are bacteria in the world. So these are literally everywhere. And they would not be able to infect my bacteria. The reason is that my bacteria wouldn't be able to read the genetic code of the viruses. So the way that viruses replicate, they don't have their own machinery.
- [40:41]
B
Oh, right, I forgot this about viruses. Okay, yeah, sorry, continue.
- [40:45]
A
So they don't have. Exactly. They don't have their own machinery to be able to replicate. They have to use the host cells machinery. And so firstly, they'd probably struggle to even bind to the surface of the mirror bacterium. But let's say they could and they inject their genetic material into the mirror bacterium. That genetic material won't be read by the host cell. So the ribosome in the mirror bacterium won't translate any RNA into proteins or the transcription of the DNA won't work. So they would be completely immune to this. And this is something we can be very, very confident in. It's basically 100%.
- [41:22]
B
Oh, my God. Whoa.
- [41:24]
A
Yeah. So why that's important is because this is a really common source of death for bacteria. So if they aren't going to be subject to this, they're getting a massive fitness advantage. And that means they might actually be able to spread in the environment and out compete other bacterial species as well. So they might be able to grow in soil or they might be able to grow in the oceans, as well as infecting all of these different species that we talked about already.
- [41:52]
B
Okay, so that's viruses and bacteriophage. Are there any predators that would be able to successfully harm mirobacteria?
- [42:02]
A
So we can be very confident that bacteriophage won't work there are a bunch of other predators that normally eat up bacteria, too. These include amoebae, or protists, which basically work like macrophages in the immune system. They engulf their prey. And many mechanisms in that process of engulfing prey and then killing it depend on interactions that probably wouldn't work with myrobacteria, which they do. Classic, right?
- [42:36]
B
Yeah.
- [42:37]
A
So I think a lot of protists are unlikely to want to consume my bacteria. Even if they could, it's not clear they're going to get much nutritional value from them. So they might actually evolve away from them rather than towards wanting to do this in the first instance. So major sources of predation, including infection by viruses, but also consumption by protists or amoebae, are probably not going to happen.
- [43:02]
B
Okay. I guess in vertebrates, our lifespans mean that we evolve very slowly. Given that some of these kind of predators probably have much, much shorter lifespans and reproduce much more frequently, should we expect some of them to evolve to be able to consume mere bacteria? Especially given that it sounds like Myobacteria could become very, very kind of abundant in the environment because of this fitness advantage?
- [43:40]
A
I think eventually, yes, that would happen.
- [43:43]
B
But it would still take a long time.
- [43:44]
A
It would probably still take a long time. So bacteriophage are effectively never going to evolve to be able to do this, because they'd have to switch their whole genetic code around, and that's a massive evolutionary step. But other predators, like protists, could eventually evolve to do this. I think it's actually more likely, and I should say this is an area where the dynamics would be so complex that it's difficult to reason confidently about it. But I think we can make some guesses. Protists initially might actually evolve to not want to consume myrobacteria when they're only present at low concentrations in the environment, because they're probably not going to get much nutritional value from them initially. So the first few protists that accidentally eat up myrobacteria, the myobacteria, might end up being toxic to the protist, and so then you'd be creating evolutionary pressure not to eat them. Once myrobacteria got to a relatively high concentration in the environment, then there would start to be evolutionary pressure for things to eat it. But that means my bacteria are already present at quite a high concentration. And I think something that's important to underscore is I don't think that myrobacteria are going to take over the whole World and out compete all other species. They only need to be present in the environment at a relatively low level in order to cause these massive risks. So you inhale about a million bacteria per day, and even if 1% of those were mirror bacteria, you'd be inhaling 10,000 my bacteria per day. 1% is about the prevalence of some of the more common bacteria that we have in the environment. So it's not crazy to imagine a bacterial species getting to that, but it really doesn't need them to be 100% of all bacteria for these risks to be the case.
- [45:32]
B
Okay. Another thing that you mentioned was changing the kind of environmental nutrient cycles. Do you mean things like the nitrogen cycle, the carbon cycle, and. Yeah, if that's right. How would that work?
- [45:46]
A
Yeah, I think, to be clear, I'm less worried about this scenario because I think it would play out over a much longer time period, and so we'd have more time to deal with it. But there are still pretty interesting scenarios to get into. We've mostly been imagining a mirror of something like an E. Coli, which needs to eat food to grow. And that food is often chiral, so that gives some limitation on its growth. But there are bacteria that don't require any food to grow. So marine cyanobacteria are an example here. These fix carbon directly from sunlight, and they can grow with just achiral nutrients like nitrogen, phosphorus, sulfur, that sort of thing. One of the very common causes of death for these marine cyanobacteria is infection by phage. So up to 50% of marine cyanobacteria die from phage infection. So if you introduced a mirror cyanobacterium, it would have basically these huge advantages because it wouldn't be able to be infected by phage, but potentially none of the limitations on nutrient availability. And so the population of marine mirror cyanobacteria might grow to be very large. If they're going to be very large and nothing is eating them because they don't really give any nutritional value to the protists that usually consume them, then they might start to sink to the bottom of the ocean and fix more carbon. It's really unclear what would end up being limiting here, so I'm not confident at all in what would happen. But it could be the case that they act as a massive carbon sink, taking a lot of CO2 out of the atmosphere.
- [47:34]
B
Yeah, okay. And initially we might be like, great, we have too much carbon in the atmosphere, but how far could it go and what would the implications be if it went Kind of beyond helping us with climate change a bit.
- [47:52]
A
Yeah. To me, it seems really difficult to kind of titrate the amount of CO2 that you'd be able to take out of the atmosphere in this way, because once you had this population of Mirastanobacteria growing, they would be really difficult to control and they would start to evolve into other things. So you might think, couldn't we just control them with mirror phages that we make or something like that? But at that stage, there'll already be a lot of them around. It'll be impossible to drive the population to zero. So we might just overshoot by miles and end up in an ice age or something like that. Equally, that might not happen, and we might end up, I don't know, killing loads of trees through mirror bacteria and releasing loads of carbon into the atmosphere. So depending on the Myobacterium that's being made, it could kind of go in different directions.
- [48:40]
B
Okay, okay, so we have my bacteria. I guess in the scenario that we've mostly been talking about, it's kind of deliberately made and released by probably malicious actors, but I guess it could be accidental by industry. If they've been using mirror bacteria for some beneficial scientific purpose, I guess it seems like that would be kind of one or two or a handful of species. How quickly would we expect these species to properly colonize the globe? Even if maybe it's not like, maybe they're not as common as the most common bacteria, but how long will it take for them to be everywhere?
- [49:32]
A
It's hard to be confident, but we can look at some existing diseases and how quickly they've been able to spread, as examples. So Covid, because people spread it through, air travel, was on all inhabited continents within four months of the first case. And more generally, the speed at which a pathogen is going to spread is dependent on how quickly the fastest moving host moves around. So if humans are being infected, it's very likely going to be humans. But insects can also spread diseases very quickly. So myxomatosis, which is a virus that infects rabbits, is spread by fleas and mosquitoes, and that was spread in the 1950s around Europe at a rate of about 7,000 kilometers per year. And the circumference of the earth is about 40,000km for context. So even if humans weren't infected, which we expect it would be, this could still travel quite quickly. But even if it's traveling relatively slowly, it's going to be almost impossible to eradicate. So it seems like once you've introduced it, it's pretty much irreversible.
- [50:41]
B
I guess the thing that really strikes me about this is I think when I originally heard about mirror bacteria, I was kind of imagining something like all of the mirror life outcompetes the kind of analogous life and like some other life forms and then we kind of end up with this weird like alternate universe, mirror everything. But it's not that. It's like mirror bacteria or kind of whatever. Mirror life ends up being released, outcompetes a bunch of other species and infects and kills a bunch of other species ranging from plants, other bacteria, non human animals and humans. And we don't get kind of, we don't get mirror giraffes maybe this is like a really trivial point, but like we just end up in a world with mirror bacteria and the like random few species that had some natural immunity. And then it stays like that until evolution works over millennia to like, I don't know, advance some life forms in some way, which is just, yeah, I mean like the definition of catastrophic, which, yeah, I think, I don't know why that wasn't like the immediately obvious implication to me, but it's just really, it's really bad. It's really, really bad.
- [52:14]
A
Yeah, it definitely is quite a bleak picture.
- [52:17]
B
Okay, I'm interested in coming back to kind of the feasibility of making mirror life or mirror bacteria specifically. So you've already kind of alluded to some of the reasons why at least malicious actors in particular might want to build release mirror life. But can you talk more about, yeah, can we do this yet? Who can do it, how far away we are from doing it, and maybe just a bit more on why different groups of people might be interested in doing this, given how horrific the implications sound.
- [52:54]
A
So ways that it could happen. Someone could invest a lot of money into it, for example, in industry if they wanted to develop this to make mirror molecules, which I really don't think would be worthwhile or good to be clear, but it's something we could imagine happening. Malicious actor might want to do this if they're trying to cause a huge amount of harm. And then a rogue AI might want to do this if they were trying to kill a huge number of people. So people in the AI community are taking this seriously as a threat that could be accelerated by AI. I think it's likely that transformative AI could accelerate the development of Myobacteria. But it's worth noting that the steps to go from where we are today to making mirror bacteria are quite wet lab intensive. They're all trial and error kind of projects. So I think autonomous scientific AI agents might be helpful here, but I don't think design is a key constraint, and I think that is an area where AI might be most helpful. So actually, relative to other bio threats, we might get less uplift on the development of myrobacteria, but it still could accelerate it.
- [54:06]
B
Okay, we can't make my bacteria yet. How difficult is it going to be? Kind of concretely, you've said it might be 10 to 30 years away, I guess 2 George Church thinks, which is insane and very upsetting. But, yeah, I guess what are the steps? And which steps do we know how to do? And which steps do we not know how to do?
- [54:32]
A
Yeah, the most likely way that I think we would make mirror bacteria is through something called the bottom up pathway, which basically means taking all of the mirror components and then putting them together and booting up life. And so there are two relevant research fields here. One is mirror biochemistry, which means making the mirror components. And then one is synthetic cells of natural chirality, because that gives you the method to boot the cell or to make life. And you need advances in both of these to be able to make mirror life. On the mirror biochemistry side, one of the key milestones that we haven't yet achieved is a mirror ribosome. The ribosome is the most complicated molecule in the cell. It's made up of 54 proteins, in bacteria, three big RNA molecules that all have to bind together in a really precise way in order to work. And no one has yet made that. The reason why making mirror components is so hard is because the way we do this normally in natural chirality is we use life to help us make things. We don't have that. So we have to make everything through chemistry. And so it's much more difficult. It's much more difficult to do it. So there are big advances needed on the mirror biochemistry side of things. On the synthetic cell side of things, no one has yet taken completely artificial dead components and put them together to make life. Once you can do that, then you have a proof of principle and a method that you could apply on the mirror. But there have been really quite amazing breakthroughs on both sides that we could go into.
- [56:13]
B
Yeah, let's go through some of those. Yeah, maybe let's start with the synthetic biology side.
- [56:19]
A
Yeah, on the synthetic. So synthetic biology is like a huge field. Synthetic cells is a subset of that, and then mirror life is like a tiny, tiny, tiny subset of that. So synthetic cells, some of the most impressive breakthroughs have been from John Glass and Craig Venter, who are two really, really impressive synthetic biologists that people will probably be familiar with, who were both on the Science paper. And in 2010, they, for the first time ever, basically took a completely dead genome that they made from bottles of nucleotides, from bottles of A's, C's, G's and T's, and they transplanted that dead genome into a living cell and then took the genome that was originally in there out of that cell.
- [57:11]
B
Wow.
- [57:11]
A
And the genome was of a different. A slightly different species. So they kind of converted a bacterium from one species to another. That is wild.
- [57:23]
B
That's actually super, super wild.
- [57:25]
A
Yeah, it's crazy. The thing that it's an exotic. I think the relevance to mirror life is it kind of shows you in principle that you can take a completely dead genome and use that to make life. So you're going from a chemically synthesized, completely dead genome, putting it in a cell, and then that genome is then the only thing in the cell after a certain amount of time, that's being transcribed and translated. So it's pretty amazing. And then in 2016, they basically did a really similar experiment where they created what they call a minimal cell. The idea here is they're trying to remove all of the non essential genes from the genome. And again, they're doing this with a dead genome they make from the bottles of A's, C's, G's and T's and removing as many genes as possible from that genome and then transplanting it in. And Mycoplasma is the simplest bacterium that we know of. It has a genome that's about a million base pairs. They managed to half the size of the genome. And so this is like an example of the most engineered bacterium that we have. We understand what a lot of the genes do, whereas in most bacteria we really don't. One of the interesting things about the minimal bacterium is that it's quite feeble because a lot of the genes that they removed are needed for being able to deal with being in different environments.
- [58:49]
B
Okay, so technically it is then alive, but it does not. It's not its best self.
- [58:56]
A
Yeah, it's definitely alive. It can do stuff in the lab. But people, when they think of really highly engineered life, often think of something like this minimal cell. And so when they're thinking about mirror life, they imagine that it would be something like that, but actually it wouldn't necessarily. It could be a mirror of an already robust bacterium like Utah.
- [59:16]
B
Yeah, okay. Yeah, that does seem important. Okay, so that's synthetic cell biology. Yeah. Talk about some of the advances from this other field.
- [59:26]
A
Yeah. So mirror biochemistry is the mirror biochemistry. I guess the central dogma is kind of the hardest part of the cell to recreate and is also the one that, you know, you would need to be able to make a mirror cell. So the central dogma is all of the machinery needed to transcribe and then translate DNA into RNA and then to proteins. And people have been working on basically trying to create that in mirror image form. The part they haven't been able to do yet is the ribosome, which is the most difficult part of that, but they have been able to do other parts of it. So people have made DNA polymerases which copy DNA, replicate DNA. So that's a key part of this. They've also made RNA polymerases which transcribe DNA into RNA in mirror form. They've made the RNA polymerase in mirror form and then taken mirror DNA that's made synthetically.
- [60:26]
B
Right.
- [60:27]
A
And transcribed that into rna.
- [60:29]
B
Okay.
- [60:30]
A
So it's like, pretty amazing. And all of these have to be made using chemistry, not through life. So they're making these really complicated molecules using. Using chemistry and using clever tricks to kind of stitch them together and get them to work. Using that RNA polymerase, people have been able to make the ribosomal RNA, which by weight is 70% of the ribosome. So although we can't make the full mirror ribosome yet.
- [60:59]
B
Making a good chunk of it, in a way.
- [61:01]
A
Yeah, we are able to make a good chunk of it. The really difficult thing left to do is put all of the parts of the mirror ribosome together correctly. So, in principle, we can already make proteins that are big enough to make a whole ribosome. So we would be able to make all 54 proteins that you need if we tried hard enough, probably.
- [61:22]
B
Wow.
- [61:22]
A
But putting them together in the right order such that they fold properly.
- [61:25]
B
Right.
- [61:26]
A
Interact properly, and then we'll be able to translate RNA into proteins is. Is really difficult and not something that people have done yet.
- [61:36]
B
So if you are trying to make a mirror bacteria, do you need to make every part or can you make kind of these parts that describe and then transcribe and then cause molecules to be created in the mirror form? I guess I'm asking because it seems like extremely hard if we get really simple. Like if we imagine creating a human body that was the mirror form, somehow you could either create every organ and Frankenstein it that way, or maybe in Theory, you could make a genome that calls for. Or that means that all of the mirror cells are created, which makes all the mirror. All the organs mirrored, or something like that. Does that make sense?
- [62:32]
A
Yeah, it does make sense, and it's a really good question because a bacterium has loads of stuff in. It has loads of proteins, loads of things in its membrane. It's just very complicated. We don't know for sure what you need to boot up a cell, but it might be the case that all you need is basically the central dogma plus a genome encapsulated in a membrane, as long as you can get the central dogma to read. To read the genome. And so it might be the case that once you can make the full central dogma, you're actually really quite close to being able to make a mirror cell. Might not be the case. We don't know for sure. But that, I think, is a way that this could happen relatively quickly.
- [63:14]
B
Yeah, that hadn't occurred to me at all. Yeah, that's kind of one breakthrough in, I guess, mirror biology. Are there others worth talking about?
- [63:25]
A
It's probably worth saying that we've been able to make mirror proteins and mirror DNA and mirror RNA for quite a long time. The challenge has been making big proteins, and we're getting better at doing that.
- [63:43]
B
Is there something like when you describe the problem of making big enough proteins that fold the way you need them to fold, that are quite complex, that does seem like a thing that AI would help quite a lot with. Is that true?
- [63:58]
A
I think it's more likely that this is a trial and error type problem where you need to be doing a lot of experimentation. So AI could be helpful if it's able to do a lot of this troubleshooting directly. But to do that, you're going to need quite a lot of wet lab footprint, which doesn't make it impossible, but means that I think it would be more difficult than a design task. There are actually other pathways through which mirror life could plausibly be created. So there's one called the stepwise conversion pathway, which would basically mean you start with a normal cell and then you engineer it so that it can create mirror components within it. So you make a second ribosome that's able to make mirror proteins and then gradually you convert it completely into a mirror cell.
- [64:50]
B
Okay.
- [64:51]
A
In that scenario, it's possible that for engineering the ribosome, it could be helpful to use biodesign tools or something. I'm still not sure that it would be necessarily that helpful.
- [65:05]
B
Okay.
- [65:06]
A
Another way AI could help is through engineering the genome to make it easy to boot. So it might be the case that some genomes are just inherently easier to boot life from. Maybe you need to start in a certain place in the genome and do everything in a particular order to get it to work. And so maybe you could be using genome design tools to think through that.
- [65:32]
B
Crazy. Okay.
- [65:33]
A
There's not going to be any data on this sort of thing. So I think we are talking about AI that would be very advanced because it's not going to have much to draw on.
- [65:41]
B
Yeah. It's going to have to, from first principles be like, okay, if we create this one molecule first, that's going to form the initial backbone that this other next one can then bind to something, something, something. Yeah. Fascinating. I'm getting a tiny bit of the flavor of this. Does sound like cool research that people are doing, but yeah, we should remind ourselves that it is not very valuable and potentially catastrophic.
- [66:16]
A
It is really cool. I'm extremely sympathetic and I think it makes complete sense that people wanted to do this. But the amazing thing is that the people who did want to do this are on the paper, come forward and said that it's not something they think should be done. So the people who were most interested in doing this stuff are now the ones that are saying that it shouldn't be done. And I think that's just a pretty amazing story.
- [66:39]
B
Yeah. Okay, let's talk about how we stop mirror life from being created kind of practically. Yeah. What kinds of governance structures or policies or treaties do you think are the right kind of approach for actually enforcing this? Yeah. This thing that all these scientists seem to agree on, which is like, we shouldn't do this, we shouldn't go there.
- [67:08]
A
Yeah. I mean, so unless new evidence comes out that really changes this picture, I think three things would need to be true in the next, say, two to five years for me to feel comfortable that the mirror life problem was basically solved. The first is that I think there needs to be a strong norm against the work to make mirror life in the scientific community. The second is that I think there needs to be regulation of enabling technologies or precursor technologies, the things that we develop on the way to making mirror life. And then the third is that governments need to be taking this seriously, such that they would deploy the kinds of capabilities that they use to stop terrorists from accessing nuclear weapons. To the question of mirror life and.
- [67:52]
B
How big are these asks, I can imagine it being the case that, you know, scientists kind of agree that, you know, we shouldn't be making mirror enzymes, that's just like, too close to mirror bacteria. But on the synthetic biology side, maybe some of the things that you'd want to kind of say, like, nope, that's getting too close, are the kinds of things that people would be really hesitant to stop work on. So I guess there are a couple of. You mentioned several different things that are kind of necessary for this to be solved. Which of these are like, yep, people will probably just say, like, yes, that's reasonable, and which seem, like, hard for people to buy into.
- [68:40]
A
Yeah. I think on the norm against mirror life, there's already been quite good progress. So UNESCO, for example, have recently put out a report from their International Bioethics Committee recommending a global moratorium on mirror cells. The UK government has looked into this and written up some notes saying that they think mirror life creation should be prevented. This German expert committee has looked into it and kind of endorsed the analysis of the risks. So I think there's already quite good momentum on that point. The second point, which is around the enabling technologies, that I think is a much more difficult decision. And the reason is because some of those technologies are things that people might want to work on in the near term. So you mentioned mirror enzymes, but actually, people can already make quite a lot of mirror enzymes, and most of that, I think, is completely fine. Yeah. There's no reason to need to go back and try to do anything about work that's already been done. But there are things that could be done in, say, the next six months or the next couple of years that some people think should not be pursued. One example there is the mirror ribosome. We already talked a bit about the mirror ribosome. It's the most complex macromolecule that you would need to make to be able to make a mirror cell. And so it's quite a natural stopping point in a way. It's also very well defined. But people in the us, people in China, people elsewhere, are interested in doing this because they think it might be helpful for manufacturing mirror therapeutics. And so there's an ongoing conversation at the moment around whether those potential benefits justify the risks of getting closer to mirror life.
- [70:25]
B
Yeah. And. Yeah. And what's the. Like, how worried are you about making just mirror ribosomes? Does that feel like. Nope. If we do that and publish papers on it that will make it too easy for, I don't know, rogue actors or something to. To do this and implement it in.
- [70:45]
A
Actual bacteria, we don't really know how far away a Miroribosome is from a mirror cell. It could turn out that there are still a lot of steps and things that make it difficult, but it also could turn out that it's quite easy. And I think it's going to be really difficult to know that in advance. It is worth saying the mirror ribosome doesn't pose any risk in and of itself. It can't self replicate, it can't evolve. It doesn't have these special risks that mirror life does have. So it's all about how much closer it gets us to making mirror life. In fact, it would be very difficult to go from there to making mirror life. Then it might make sense for it to happen. There are also even different types of mirror ribosomes. You could make a basic mirror ribosome, but a mirror ribosome that's really good at making proteins might be quite a bit harder to make than that. And that's probably the sort of thing you'd need to be able to make mirror life.
- [71:36]
B
Okay, that makes sense. Will we learn much more about what are the really hard steps here? It's interesting to me that it's hard to predict in advance. Is there any reason to think that we will actually find out it's not the mirror ribosome bit, it's like bringing all the pieces together and animating them bit, or is that just like. We probably can't know because we would like to stop before we do any experiments that shed light on that.
- [72:07]
A
I think we could learn more, especially from experiments on natural chirality cells. So there are kind of these two areas that we've talked about, mirror biochemistry and synthetic cells, both of which ultimately you need to have progress in to be able to make a mirror cell. And we might learn things from natural chirality synthetic cells that tell us either that it's very easy to go from a basic system with a membrane to life, or very difficult. But there are hard questions that need to be answered on the synthetic cell side of things too. The challenge there is synthetic cells is a much bigger field, at least by some definitions, and there's a lot more interest in pursuing that. So any kind of restriction or regulation there might impact more ongoing research. And that is a trade off, right?
- [72:54]
B
Yeah. What are the most contested debates on that side of things?
- [73:01]
A
I mean, one intuition here is that if you get good enough at making synthetic cells, if you could make a completely artificial cell in the lab, then you could use the same methods to make a mirror cell, because the chemistry would all be the same. So if this got really advanced, then you might be quite close to making mirror cells, but we're not really very close to that at the moment. In reality, when people are talking about synthetic cells, they mostly are using some parts that are derived from life. And if you're using parts that are derived from life, then you can't immediately replicate that on the mirror side of things.
- [73:38]
B
Okay. And so just to make that more concrete, it's something like, I'm sure this is not actually it, but like if you could use some subcomponent of a ribosome to create a full ribosome, that might help you a lot in synthetic biology, but it doesn't help you with mirror life because we won't be starting out with that half ribosome to then build on.
- [74:07]
A
Yeah. In mirror life, at least from the bottom up pathway, you have to make everything from scratch. And that makes it a much more difficult problem. And in some ways there's not a great reason why you'd want to make everything from scratch on the natural chirality side of things. So that helps you to some extent. But I think it is a long term goal of science to be able to make life completely artificially. And so eventually we are going to have to think about what's an appropriate line there. I think we can probably get a lot of the benefits of synthetic cells without making them completely artificially. And so I think we need to think about what lines there would neatly delineate the things that get us too close to mirror life and don't. But this is a topic that a lot of people need to weigh in on.
- [74:57]
B
Yeah. Is it like, are there arguments for doing at least that particular thing, creating a cell from total, total scratch that aren't just to do with like. It sounds really cool. And it makes sense that people would be keen to do it because it would be a really massive achievement. Like, to what extent are we giving things up by drawing lines before that?
- [75:25]
A
Yeah, I think this is mostly an area of Blue Sky's research. We've been having discussions with a lot of the people doing this work over the last few months and I haven't heard a specific benefit of completely artificial life be articulated. The idea is more that you could generally manipulate life and if you understand it really well, you can do that better. And so you could do a myriad of different things. But I do think a bit more critical thinking is needed in terms of whether you could achieve those benefits the other way. One example is you might be able to just transplant genomes that have all of the characteristics that you need into cells, and then you basically have the cell be defined by the genome instead of booting it up from all of its individual components. And that might get you basically the same benefits.
- [76:15]
B
Yep, that makes sense. For people who are going to be sceptical of governance, that stops some synthetic biology research. What are the kind of fair things that you concede, yeah, we'd be giving something up, but. And maybe you think we still should, but are there types of research that probably we shouldn't do where you can actually see a point to it that's instrumental?
- [76:51]
A
The things we've been talking about are the foreseeable benefits of making mirror life or making the technologies on the way to doing it. A lot of science is done without a specific application in mind. And usually I think that's a good thing. Like we should have a prior on scientific progress being good, given the results that it's produced historically.
- [77:10]
B
Right.
- [77:11]
A
In this case, we have to weigh that up against the. What I think is pretty overwhelming evidence of the risks, and I don't think it's sufficient. One thing that I think is really important to say is that synthetic biology as a whole is a really big field that has a huge range of benefits that I do buy into. Synthetic cells are a relatively small part of that. And within synthetic cells, I think there are a lot of applications that will ultimately be important. Mirror life and mirror biochemistry and possibly some specific synthetic cell experiments that might not make sense to do are a tiny part of the technology tree. So it's not like by not pursuing this we'll be giving up a huge area of science. This is one very tiny area of the technology tree that we could choose not to go down.
- [77:58]
B
Okay, so it just sounds like the arguments are really compelling. The benefits are not that big and are going to be reasonably easy to kind of accept. We're going to give them up. How optimistic are you that. Yeah, as you and this group kind of take more next steps, you'll get the kind of policy and governance measures you think are really important kind of passed and in place.
- [78:30]
A
I think that's a really hard question to answer. I think we'll get some better answers to some of the most important questions within the next year or so. But then it's going to be a really hard road to go from there to implementation. And a lot more people are needed to think through how exactly that will work. Each different country might have a different regulatory framework to think through something like this. And there's only a really small group thinking about this at the moment who are not going to be expert in all the things necessary. So this is one of the areas where we really need more people to come in and think about how this should go.
- [79:05]
B
How hard do you expect it to be to get these kinds of policies passed? And I guess one question that comes to mind is have we successfully implemented global regulation on a kind of scientific technology like this before?
- [79:23]
A
Yeah, we have. Actually. There are some really good examples here. Nice one is the Environmental Modification Treaty, which I didn't know about until quite recently. But basically this is a treaty that quite a lot of different countries signed on to in the late 80s, prohibiting the use of environmental modification as a weapon. So that means using weather as a weapon.
- [79:50]
B
Right. Okay, interesting.
- [79:51]
A
And the US and the USSR both signed onto this and kind of collaborated on it. It's a super interesting one that. I'd love to know a bit more about the history. But it is an example of a technology that didn't exist being prohibited well in advance. So I think this is one example.
- [80:12]
B
Yeah, nice.
- [80:13]
A
Other areas include the Montreal Protocol, which is one of the most successful examples of an international treaty prohibiting ozone depleting substances. There's also a really not well known treaty prohibiting the use of blinding weapons like lasers. Things like this have been agreed to before. So I think there is actually quite a lot of precedent showing that we can identify something that doesn't yet exist and decide not to pursue it.
- [80:44]
B
Yeah, I guess I also thought of cloning, but I wonder if cloning is a slightly mixed example because there is agreement that we shouldn't be cloning humans. But I feel like I'm remembering correctly that some people ignore this and do it anyways. Do you worry about that with Mirror Life and also maybe fact check my cloning thing?
- [81:13]
A
Yeah. So human cloning no one has done. In 1996, Dolly the Sheep was cloned. And then there was this big discussion that ultimately resulted in people agreeing that human cloning shouldn't be done. That's an interesting example where the technology is kind of in principle there to do this, but no one has actually gone ahead and done it. The thing that you're thinking of is human germline editing, which is a different technology. What that demonstrates to me is that a norm on its own that lets you go right up to the edge of being able to do something is not sufficient. Because you can't assume that people are always going to go along with a regulation or with a norm. And so that means you need to draw lines earlier on that mean that if people cross those lines, there's still space between where they are and the thing that you're worried about.
- [82:05]
B
Okay, that makes a lot of sense. Thank you. Okay. I want to come back to how people can maybe get involved and what the next steps are for this project. First, we've talked a bit about some of the countermeasures that might and might not work. But are there any kind of potential countermeasures like vaccines or something else that that you think are really promising that we haven't talked about yet?
- [82:32]
A
Yeah, I think there are. So you can use conjugate vaccines to kind of trick the body into having a robust response against basically anything. So you can use them to trick the body into making antibodies against cocaine. You could do the same potentially for mirror components. And that's something that I think is worth exploring in some more detail. More generally, physical countermeasures should be helpful against mirror bacteria. So as a kind of plan B backup if prevention doesn't work, I do think investing in robust biodefence, that's kind of threat agnostic, could be very valuable and is something that people should think about doing.
- [83:17]
B
Can you give specific examples? I'm picturing bubble boy, again, basically.
- [83:21]
A
That might be the extreme. That might be the extreme. But in the nearer term, things like ppe, bio hardening, possibly early warning systems, Some of the things that Andrew Snyder Beattie talked about on a recent podcast on 80,000 hours, I think many of those could help for a myrobacteria outbreak, too.
- [83:41]
B
Oh, yeah. I am kind of interested in early warning systems. Would we know that if we started seeing deadly infections in hospitals, they were caused by mirobacteria? Like, how difficult would it be to notice that?
- [83:58]
A
That's a really good question. A lot of the methods that we use to detect things use enzymes that wouldn't work for mirror life because they're chiral, because of the chirality of them. So PCR is an example here, which is used to kind of amplify DNA. We can actually already make all of the enzymes in PCR in the mirror form. So we could in principle use those enzymes as a kind of detection method, but at the moment, we're not set up to do that.
- [84:31]
B
Yeah. Okay. Okay. Yeah. Are there other kind of biodefensy things that you think are promising?
- [84:40]
A
I think more work on antibiotics could be good. I'm not sure whether it makes sense to develop new antibiotics or just think through the ones that already exist and figure out how best we could scale them up. But I think People thinking through that problem might be beneficial. One set of countermeasures that I don't think will work and that the world is investing a lot in are MRNA vaccines and DNA vaccines. Because the way those work is they use your cells to create proteins from them. So you put an RNA vaccine into your body and then the cells make a protein from the RNA and then you have an immune response against that.
- [85:23]
B
Right.
- [85:24]
A
If you were injecting like mirror rna, your cells can't read that genetic code, so it won't work.
- [85:32]
B
Lovely.
- [85:34]
A
So unfortunately, some of the platforms that we're kind of doing the best at are not going to work here, and different approaches I think would be needed.
- [85:41]
B
Okay, I guess those are some countermeasures that would and wouldn't work mostly in the context of humans. What kinds of things do you imagine helping? I guess I can imagine many of those things also helping with non human animals. But maybe plants are quite a different issue. And maybe in particular, thinking about countermeasures for crops is super important. Are there any that seem promising there?
- [86:08]
A
Yeah, so we could actually engineer crops to be able to detect my bacteria infections. We could probably engineer them so that some of their pattern recognition receptors were able to bind to some of the common molecules on Myobacteria. And this is something that also might be worth some further research. I think we can do that probably to protect a small number of critical crops. You're going to have to do that for basically every species that you want to protect. So it's very difficult to scale so that you could protect the Amazon rainforest and all the different plants in a forest near you. But it could be done for some key crops, I think.
- [86:47]
B
Can you help give an intuition for why that is so hard to do at scale? Intuitively, we engineer crops all the time. Why can't we just do that?
- [86:59]
A
I actually don't know how difficult it would be to do this for a crop. I think it would be great for people to think more about this. The point that I'm trying to make is that it would be impossible to do this for natural environments. So we could do it for crops and a lot of the food that we plant directly ourselves. But doing that across other environments, other plant species, seems like it would be a really difficult undertaking.
- [87:22]
B
Okay, let's say that we have a bunch of wins in these countermeasures and maybe we can't do a bunch of them at scale. What is the kind of best realistic outcome? Yeah, it seems like we shouldn't assume that we'll be Able to get antibiotics to everyone who needs them, or design kind of every crop to be immune to mere bacteria. So what should we be picturing in a world where we technically succeed? It still doesn't sound that good.
- [88:02]
A
Yeah, I think this is a really hard question, but I think we could protect at least relatively small populations of humans and enough food for us to be able to survive as a species. Beyond that, I think it's really hard to say. I mean, hopefully we just don't get to this point and we focus on prevention, which I do think is much more important for most people. But I think we can protect against the very worst harms. And so people working on this and thinking through some of these measures in more detail, I think could make a really big difference.
- [88:37]
B
Nice. You've spoken to, I think, hundreds of people about this issue, and my sense is that you tend to get lots of just kind of questions and probably a fair bit of skepticism, at least at first. So I'm interested in going through some of those common questions and your responses to them. The first one is it seems like mirror bacteria, maybe mirror life more broadly, would have this very significant competitive advantage relative to normal life. Given that, why do we need to engineer it synthetically? Why wouldn't it just arise on its own through evolutionary processes?
- [89:19]
A
Yeah, it's a good question. And the answer is that there's no stepwise process through which mirror life could evolve from normal life that would be advantageous to the evolving organism throughout that process. So normal life is on one fitness peak. Mirror life could be on another fitness peak. There's a valley that you have to cross in between, and all of the steps that you go through would be harmful to the organism. So if you try and incorporate mirror amino acids into normal proteins, it makes them fold. It makes them fold in weird ways and not work properly. So the whole organism is going to kind of break when this stuff happens. One good analogy, I think is driving. So if you have a road, you have people all driving on the right hand side of the road to swap over. You can't just have one car at a time move. That will just really get you in trouble in the traffic. You need everyone to switch at the same time to get any advantage there.
- [90:19]
B
Yeah, okay. Yeah, that's a great example. What about. So we have what we consider normal life, which is made up of molecules in a particular arrangement. Why didn't kind of mere life evolve just totally independently? So not from current organisms, but kind of beginning of the tree of life? Why didn't we get mere organisms that could kind of from first principles, end up being fully fledged organisms that. That then outcompete normal life.
- [90:56]
A
Yeah, I mean, no one really knows how common it was for life to originate for this process of abiogenesis to happen. So it might be the case that it's just incredibly rare and the life that we have was really unlikely to happen, and it happened once and there just weren't other opportunities for mirror life to evolve. If you think that abiogenesis is more common, then an answer here is that once any type of life was evolving, it was gaining a lot of fitness, advantage and ability to use resources. So any new life was going to be really feeble in comparison to it. And when we're talking about engineering mirror life, we're not talking about starting with a really feeble life form, which is what you'd have in the kind of primordial soup. We're talking about going across this full fitness to a new fitness peak completely in one go. And so you don't have that disadvantage.
- [91:49]
B
Okay, yeah, yeah, that makes sense. That makes tons of sense. Pushing on then, another question I think you get quite a lot is we have this part of our immune system that can create antibodies that can bind to anything. So part of the immune system is specific to things that we've kind of seen historically, and so it's ready to react to those things, but part of it is much more adaptive. Why isn't that enough to mount effective immune responses?
- [92:21]
A
Yeah, so your body probably has antibodies that could in theory bind to mirror proteins, but it would be only making them in really small quantities. The issue with mirror life is that it wouldn't activate your antibody factories, your B cells, such that you produce enough antibodies for it to be able to kill the invading bacterium. And that's because the innate immune system is probably not going to be activated properly. But also parts of the adaptive immune system that kind of downstream of this binding are not going to work.
- [92:54]
B
Yeah, I feel like there's this meta point here that has come up a few times where you can have a simplistic understanding of different parts of the immune system that might feel like they'd be reassuring like we have. I mean, this is a clear example. We can make antibodies that bind to anything.
- [93:19]
A
But.
- [93:20]
B
But when you actually look at the science and realize that these different components of the immune system are incredibly interdependent, you realize why we can't just count at least most likely on one part of the immune system being, I guess, less sensitive to chirality or Something.
- [93:43]
A
Yeah, I think that's basically right. The immune system has a lot of interdependency, but also redundancy. Your immune system can actually be really dangerous to you. So usually for things to be activated you need more than one signal and that's so that your immune system doesn't attack itself.
- [93:59]
B
Interesting.
- [94:00]
A
Yeah. A lot of common diseases are the result of the immune system going wrong and actually attacking you, like autoimmune diseases. So there's a lot of regulation and a lot of redundancy in the immune system. And that means that often you need more than one signal to get things started. And we know that many of the signals are not likely to work.
- [94:19]
B
Yep. Okay. Okay. I think that's helpful because I think it's just like I can imagine it being really tempting to be like, haha, there's this one part of the immune system that's totally going to work against mirror life. And then yeah, you're reminded of these facts that mean that that part just wouldn't be activated at all. And that just seems, yeah, really important, potentially non obvious.
- [94:40]
A
When I speak to my immunologist colleagues about this, they said when they first heard about this issue, their initial reaction is to kind of defend the immune system. Because the immune system is amazing. It can do so many things and it is really adaptive, but it also is very complex. And breaking one part of it we know can cause really big problems. So we can't be 100% sure about anything that will happen here. But the fact that we can be pretty confident that certain key parts of the immune system won't work, I think is quite worrying.
- [95:12]
B
Yep, nice. Okay, that's really helpful. I think another question you get fairly often, which yeah, I think I had. We make antibiotics all the time. It sounds like in theory kind of analogous to making antibodies that can respond to mirror bacteria. We should be able to make antibiotics that have the right properties to respond to mirror bacteria. Why wouldn't those be enough?
- [95:41]
A
Yeah, I mean a lot of antibiotics are chiral and so they wouldn't work. But some are achiral and some come in racemic mixtures which include both chiralities. And so those would be expected to work. I think the key difficulty with any medical countermeasure is scaling it up and distributing it to all of the people that would need it. And even more than that, giving it to all the animals and plants that could be infected. And you'd need different countermeasures, most likely for each of the different species that could be infected. So this would be doing something at a scale that we really can't do at the moment and historically haven't been able to do.
- [96:21]
B
Yeah, I guess if I'm trying to wear the hat of the skeptic who's like, I don't think that this risk is actually that bad, maybe I hear that and think something like, okay, there is a solution. You're saying scale up would be hard, but it sounds like antibiotics would work in theory. Given the stakes, surely we would figure out how to scale this up. Why don't you buy that?
- [96:51]
A
I think it's possible. I mean, I do think this is something people should be thinking about and should be working on, particularly if you care about reducing X risk specifically or you care about longtermism. I think there's quite a good argument to think that by investing in countermeasures, you can cut off some of the worst risks from mirror bacteria. I'm just not that optimistic that society will be able to get itself together to do this at a massive scale, given our track record. But that doesn't mean we shouldn't try. And I think it would be great to have people thinking about what that would need to look like.
- [97:20]
B
Yeah. I have this feeling that there's something that I'm not totally understanding about why it would be so difficult to not just scale up globally, but to scale to different species. Is there something you can say there to make it more intuitive about? Because. Yeah, where is that coming from? I think something like, I don't know, lots of antibiotics that work on humans also work on dogs. Why do we need to invent a bunch of different things for this to work out?
- [97:53]
A
Yeah, I mean, I think a lot of this is about getting it physically to the place where it would be needed. So if you imagine that you want to try and protect, like the rainforest or various wild animals or all different livestock, you have to physically get the antibiotics to all of these species. The other thing is just giving it to them once is not enough. They have to receive it continuously, because if my bacteria are in the environment, you could get infected directly from the environment. So they'd need to be taking antibiotics every day to prevent them from getting infected in the first place. And that would mean you need to make loads of it, basically. And so we might need just much more manufacturing capacity than we actually have.
- [98:37]
B
Yeah, actually, I want to understand that better. I don't think I fully understand why we need to be taking it continuously. You mentioned it before that people with similar immunological deficiencies have to be on Antibiotics, basically forever. What is going on there?
- [98:57]
A
I think the basic idea is that you need to try and avoid getting infected in the first place because your immune system is so bad at dealing with it that once you're infected, it's really hard to clear the infection. Antibiotics help you go below a threshold after which your immune system can take over and do the job of clearing the infection. But that might not work in this case.
- [99:22]
B
I think I want to come back to countermeasures in a bit. But first, yeah, talk about another question I think you get sometimes. So mirror life won't have evolved in the environment in the way that normal life and bacteria in particular have evolved in the environment. You might think that they've kind of their structure is adapted to not be harmed by, I don't know, particular compounds that are in the environment, but maybe mirror life will be kind of vulnerable to those kinds of compounds or to some kinds of compounds. Is this a worry? Like, will mirror life just find the environment toxic?
- [100:09]
A
Yeah, I think it's a good question. So one way we could understand this is by making mirror life and then putting it in an environment and seeing if the environment is toxic to it. Another way you can study that is by looking at normal life and kind of putting it in a mirror environment and you get exactly the same information. We actually already have some data on that that shows that at high concentrations, mirror compounds will be toxic to normal life, and so the reverse would be true. But those concentrations are higher than what you typically find in the environment or in the blood, the places where you'd expect mirror bacteria to be growing. So there are more experiments that could be done here. I think it could be great for people to be making more mirror environments, putting bacteria in them, seeing how they survive. But the data that we have at the moment suggests that this is not going to be a problem.
- [101:01]
B
Okay, that's really cool. This, as I'm asking, I'm seeing problems with it, but I'm going to ask anyways. If we learned that there was some particular compound in the environment at some level that wasn't quite toxic, but if you turned it up 1%, it was then very toxic to Mary life, would that be a countermeasure you'd be interested in, or is that in the, I don't know, realm of no, we shouldn't turn compounds up by 1% in the environment because we don't know what's going to happen?
- [101:39]
A
I think it's possible. I mean, if there was a myrobacteria outbreak, then I'm sure it would make sense to try stuff like this. I don't think it would be a very good preventive measure because I'm sure there would be other impacts that we haven't thought about of changing the whole of the world's ecosystem in one go. And I actually think there are probably more promising countermeasures that that would kind of spread on their own that might be worth talking about more.
- [102:02]
B
Huh. Like what?
- [102:03]
A
So mirror phages are one example. We talked about phages already, which are the viruses that infect bacteria. Normal phage wouldn't be able to infect mirror bacteria because the mirror bacteria can't read the genetic code of the phage. But you could synthetically make a mirror phage or a mirror virus that would be able to infect mirror bacteria. Phage are one of the main things in the wild that kill bacteria and they replicate when they kill the bacteria. So you could use this as a way to control the population of mirror bacteria to some extent. The difficulty is that you can't use them to completely wipe out a mirror bacteria population because you need a certain density of the host population for the phage to be able to replicate.
- [102:53]
B
Hmm. So something like, is it just kind of this fact about kind of populations and ecosystems where, like, you can never. Like a predator could never, or like, it's difficult to completely eradicate a prey species because once there are just a few, the predator won't be able to survive and kind of flourish on just those few. And so their population will be knocked down and then the prey species will bounce back.
- [103:28]
A
Exactly. You end up getting these kind of population dynamics where the prey population will drop, that then means that the predator population drops, which then means that the prey population can expand again, which then causes the predator's population to expand. And so you end up with this kind of like, up and down trend. But it's very difficult to completely eliminate a host species in this way. And we've never been able to use a biological control to completely make extinct a species that we've tried to do this with. It could help a bit, but you're still in a pretty bad situation there, because you have my bacteria in the environment, pretty much impossible to get rid of them completely. And so you have this transmission route from the environment to multicellular species.
- [104:15]
B
Yeah, yeah. That seems pretty terrible. I guess I can imagine. One question you might get quite a lot is so much of what we've been talking about has uncertainty so exactly how the immune system would fare, whether some species would be Randomly actually well suited to kind of developing immune responses to mere bacteria, I think. Yeah, there are loads of others. How easy would it be to get antibiotics that work kind of across a bunch of different species? Given all of that, do you endorse trying to take a bunch of action on this now as opposed to maybe doing a bunch more experiments before we spend loads of resources trying to make mirror bacteria never exist?
- [105:16]
A
I do think it's important to do things now, and I think that comes from the fact that there's kind of an asymmetry here where we can pause on work towards making mirror life. And if we learn more, we can always unpause and continue to do that work. If we've already done the enabling work to making mirror life and we learn that it is, in fact, really dangerous, we can't undo that work. That knowledge is already out there. And so I think it's much more prudent to take action now and then undo it if we learn something else. But I also think there is in fact, a lot of evidence already that this is a problem. The technical report, which is this 200 page document that many of the top scientists contributed to, goes through all of the evidence that's available now. Most of it wasn't generated specifically trying to assess the risks of mirror life, but it was done and we can use that evidence to think about the risks. So there are more experiments that could be done, but there's actually a lot of literature already that we're drawing on to come to these conclusions. One thing I want to say up front, though, is that because with mirror bacteria, we're not just talking about a single species, we're talking about a whole class of organisms that could include many different species, kind of like all viruses or something. It's very difficult to do experiments that would let us rule out the risk. We can do experiments that would help us to learn more. Might be helpful for thinking through countermeasures and things like that. But I'm not optimistic that in the near term there's a small number of experiments that we could do that would shift the risk benefit balance into the other direction.
- [106:55]
B
Yep, Yep. Okay, moving on. A thing that we kind of alluded to a few times but haven't really fleshed out. You've kind of said a few times the benefits aren't that compelling here. But yeah, maybe we should talk about what, if any, benefits there are. So in theory, aside from the fact that it would be very cool, why do people want to make mirror life besides this kind of rogue Actor, malicious actor thing.
- [107:26]
A
Yeah. I mean, I think now no well meaning scientists want to do this, but there were reasons that people want to do it before. I do talk to a lot of people about this who think we should reject the kind of risk benefit framing for an issue like this where you have catastrophic risks that could kill a large fraction of multicellular life. They'll ask, okay, what is the benefit that you can actually imagine that would justify doing that? And I'm somewhat sympathetic to this, but I think it is also important to get into the benefits to some extent. And in this case, I don't think it ultimately makes any difference in terms of what we should do. So what are the benefits? The key benefit that people talk about for making mirror life would be making mirror molecules which can be used for therapeutic purposes more efficiently. But we can already make a lot of these mirror molecules without mirror life. So it's specifically focused on large mirror molecules.
- [108:27]
B
Okay. Yeah. So I'm trying to generate some hypothesis for why having big mirror molecules is medically useful. And I'm not coming up with anything. What are the kind of applications?
- [108:42]
A
Yeah, so the properties of mirror molecules in general that make them interesting for therapeutics are the fact that they tend to not produce an immune response and they don't get degraded in the body, so they can last for longer. But those are the exact same properties that make mirror life dangerous. Having said that, mirror molecules, there are already quite a lot of these in use, at least in early development. So this is an area where there's some investment and some interest. Big mirror molecules, it's a little bit unclear to me what exactly you'd want to use them for. Maybe if you wanted to make a mirror enzyme that could cut something up, then you might want to do it. Most of the mirror molecules at the moment work by binding to something, so they don't need to be that big and complicated. If you want a molecule to have lots of. If you want it to have a complex function, then it needs to be a complex molecule that can kind of move around and have different pockets and things. And so maybe there would be applications for those mirror enzymes, but I'm not really sure what they would be.
- [109:53]
B
Okay, and where is the line that worries you? I guess. Do you feel fine about mirror enzymes if we do find uses for them? Or are you like, nope, that is too close to mirror bacteria?
- [110:07]
A
No, I think all the work that's happened already for therapeutic purposes doesn't pose any special risk. It poses the same risks as any new drug development. So Might have things that you didn't predict would happen, but it can't self replicate. And so I don't think it's particularly dangerous. Where the line should be between those and mirror life is a really difficult question. And one of the key open questions, I think, at the moment in the discussion around mirror life that we'll probably talk about in a minute.
- [110:39]
B
Okay, what about the argument that we should basically make mirror life so that we can make countermeasures to mirror life, which I think is a pretty common argument for other types of biological threats. We want to make certain kinds of new, scary pathogens so that we can develop vaccines for those pathogens. Yeah. Do you find that compelling at all?
- [111:03]
A
I think the risks from making mirror life would just be far greater than the benefits that you'd get from being able to develop better countermeasures. But even more than that, we've already talked about how mirror experiments can tell you a lot about mirror life. So if you want to develop a countermeasure against mirror life, one way to do that is to make mirror life. And this is the way that I don't endorse, is to make mirror life and then test countermeasures against it. The other thing that you could do is use normal life as a model organism and make mirror countermeasures against normal life. That means if there was ultimately a mirror life outbreak, you could just take the mirror of the countermeasures that you've developed and scale that up instead.
- [111:49]
B
That's incredible. I feel like, yeah, for the most part, all these properties of mirror life seem terrible and make this a really upsetting, scary issue. And then occasionally there are properties of mirror life that are like, actually this makes developing countermeasures much, much safer to do than normal pathogens, which seems just like something to be very grateful for in a way.
- [112:14]
A
We know a lot about mirror bacteria and how they would behave because physics tells us exactly what would happen. It would behave in the same way as normal life in isolation. And that's a really unusual property. We can't usually reason so effectively about something that doesn't yet exist. But I think that reduces the benefit of going ahead and making it. And some people think that making mirror life is just a vanity project because we know exactly how it would behave.
- [112:45]
B
Okay. Another kind of fun question that I think you get at least sometimes is, I guess you can imagine we've already talked about why you wouldn't get mirror bacteria, for example, evolving randomly on Earth now. It just like it wouldn't outcompete normal bacteria if it had to start again. But you can imagine on other planets the kind of the mirrors of our life evolving because that's just like randomly what came about from kind of abiogenesis when life evolved on those planets. And maybe you think that we're eventually going to travel to those planets, which would mean being exposed to those mirror forms of life. Is there an argument that we should create mirror life now, learn really good countermeasures? Because otherwise we could be exposed to it on another planet unexpectedly and not have any countermeasures then?
- [113:53]
A
No, I don't think so. But the reason for that is the type of mirror life that we're talking about here that all the analysis that's been done is focused on is an exact mirror of life on Earth. It's not something that would have some mirror components and look a bit like what we have on life on Earth. Because the special risks come from the fact that mirror life is already well adapted to living on Earth. Because it gets to benefit from all this evolutionary history. It's vanishingly unlikely that an exact mirror of life on Earth would evolve on a different planet. So even if we rewound the tape of life on Earth, we probably wouldn't get something that looks exactly like what we have today. And so doing that on a different planet, it's just incredibly unlikely that there wouldn't be some contingencies that pushed life to evolve in a way that's a bit different.
- [114:45]
B
Right. You're not going to go to planet X and find mirror E. Coli. It's just extremely, extremely unlikely.
- [114:54]
A
Yeah, exactly.
- [114:55]
B
Makes sense. I'm kind of interested in the process of bringing all of these scientists together, having this kind of consensus emerge and then writing this paper and releasing it. Can you kind of talk about what some of the early conversations were like?
- [115:16]
A
Yeah, for sure. When I first got involved in this, which was about the middle of 2024, this working group was just starting to form that Jack Szostak and John Glass were co chairing. So I worked really closely with them and a few others to basically write up the paper, recruit more people into the working group and think about how we wanted to communicate this to the world. So we had these meetings with all of these top scientists from China, from Japan, from the us, from Europe, where we'd all be on zoom calls, kind of going through the arguments, going through the exact recommendations that we wanted to make in the paper. It was a pretty crazy time.
- [115:56]
B
That is crazy. Yeah. I didn't realize quite how International. It was, yeah. It sounds like this is just really scientists everywhere.
- [116:06]
A
Yeah, I mean, it was really international. So some of the top synthetic biologists in China were involved in the group. Someone called Chen Li Liu, who is on MBDF's advisory board. He runs like a 6,000 person synthetic biology institute. Hiroaki Suga, who is one of the top synthetic biologists in Japan, who's founded big successful companies. So this was also a group that is not naive to the. The potential benefits of new technologies. These people, a lot of them have been successful in biotech and are really pushing the frontier of synthetic biology.
- [116:42]
B
Right, right. Okay. So that was kind of the beginning. Were there, I don't know, particularly big hurdles or particularly big wins from there?
- [116:52]
A
One thing I'll say is it was quite hard to get 38 people to agree on the exact text that we used in the science paper. We literally had these Nobel Prize winners, all these top scientists from all these different countries in a single Google Doc, commenting back and forth, arguing about different bits, trying to agree on the text. So that was pretty fun.
- [117:13]
B
Wow. Yeah, I mean, I have quite a bit of imposter syndrome. Did you have any of this working with. I mean, you're an accomplished scientist, but working with a bunch of Nobel Prize winners?
- [117:27]
A
It was definitely a bit intimidating getting introduced to all of them and kind of. When I first joined, I flew around the US to meet some of the key people, like John Glass and David Relman, met them in person, talked about synthetic biology, which honestly I didn't know that much about at all at the time. At this stage. I've spent a fair amount of time thinking about this, but I was new to Mirror Life and I couldn't believe that it was even something we might be able to realistically do in 10 years. I hadn't been tracking a lot of the developments going on in the synthetic biology field and learning about them. I was like, whoa, this is way further along than I would have guessed.
- [118:03]
B
Wow. Yeah. Was any of it particularly stressful or shitty from a personal perspective?
- [118:13]
A
It was quite hard work there, was it?
- [118:17]
B
Imagine a really busy time.
- [118:19]
A
It was a really busy time, but it was also amazing to get to work with such great people and people who fundamentally cared about what we were doing. These are people who previously, in a lot of cases, wanted to make Mirror Life and now were coming forward and saying, this is something that shouldn't be done. So although there were a lot of hours, there were a lot of phone calls with many different people from all around the world. It was really motivating to be a part of that group and to know that the way we talked about this was going to make a really big difference to how the issue went. So I was also pretty nervous because we had to make some decisions when we were writing up the paper about what recommendations we'd make. This for many people was kind of coming out of nowhere and there's a different world where it could have been perceived as this crazy sci fi risk and not taken seriously. And it was really important to avoid that from happening. So when the paper finally went up, we were kind of holding our breath to see how it would go. And I think fortunately it seems to have played out pretty well since then.
- [119:24]
B
Nice. How did you and this group navigate this worry about information hazards and not wanting to tell potential omnicidal actors that Mirror Life is this great way to cause the extinction of a bunch of species, including maybe humans?
- [119:45]
A
Yeah, I mean, we thought a lot about this and there's a section in the technical report at the front which I think does a really nice job of summarising the top line thinking. But the main argument that I find most compelling is that by default we were on track to make Mirror Life eventually. So a lot of well meaning researchers had said this is something they wanted to do. It had been this kind of long term goal of synthetic biology, this interesting thing to aim towards for quite a long time. And so by publishing the paper and drawing attention to the risks, I think we were letting the conversation about what should be done happen much earlier than it otherwise might have. By default, people would have continued developing technologies relevant to Mirror Life and then either we'd have just developed Mirror Life, which could have been really bad, or people would have spotted the risks nearer to that time and then we'd have less time to deal with them. So I think that for me, I found very compelling.
- [120:46]
B
Yeah. Yeah, that makes sense. One very striking thing that we've already talked about a little bit, but that I'm kind of curious to hear more about, is this fact that some of the scientists who ended up getting really on board and wanting to talk about these risks openly were also the scientists who were really excited about this research and kind of working on it. Do you have a sense of what their journeys were like and how difficult it was for them to move from one side to the other?
- [121:21]
A
Yeah, I mean, Kate Ademala, I think, has been an amazing person here. She was on the National Science foundation grant in 2019, which was specifically about trying to make miracelles. There are a Few other people on the paper from that grant too, and I think. I don't want to speak for them, but I think the arguments around the risks just basically convince them. Apparently when this was initially being discussed, John Glass was part of the NSF workshop where the idea came up and he said everyone talked about it, they just thought, yeah, sounds like a cool idea, let's do it. But there were no immunologists in the room, there were no ecologists in the room. And I think that interdisciplinary group was what was really needed in the end to fully understand the risks. The risks have been alluded to in different bits of literature over the past 20 or 30 years, but there was a bit of a disconnect in that no one had looked into them deeply and realised that this could actually be really bad. People had kind of mentioned it in offhand comments. So something that I've taken away from this is you do need to take really weird ideas seriously sometimes, but you also need to do proper due diligence on them, because stuff like this, the vast majority of the time will turn out to not be a problem, but the one in a hundred times when it is, that's really important to realise.
- [122:43]
B
Totally. Okay. For people who are listening, who feel compelled and convinced, and maybe feel like they might be able to contribute, maybe to give some context, what is the kind of state of where this project is? What are some of the current things? What are some recent accomplishments? What's happened since the paper came out?
- [123:05]
A
I think there's really good momentum around this being taken seriously in the scientific and policy community. So in February, there was a meeting that was the 50th anniversary of the 1975 Asilomar Conference on Recombinant DNA, where they first kind of proposed a moratorium on recombinant DNA. Out of that, a statement signed by nearly 100 people from all over the world, including China, the US, Europe. I think actually someone from every inhabited continent signed on, said that they thought miralife should not be created if these risks are as bad as they seem. The UK government has looked into this. A German Biological Safety committee has looked into this issue and reaffirmed the key assessments. And UNESCO has an international bioethics committee which has recently put out a report recommending a precautionary global moratorium on making mirror cells. Yeah, so I think there's been a lot of progress around, let's not do the final step of making mirror life. And the big open questions are now kind of, where should we stop on the way? And that's something that we're working on a lot at mbdf. Nice.
- [124:16]
B
Okay. And so for the people who are thinking they might want to get involved, what kinds of skill sets are you most interested in?
- [124:26]
A
One of the amazing things about working on Mirror Life is the number of different disciplines that it touches. So over the weekend, preparing for this interview, I was brushing up on ecology, synthetic biology, immunology, policy, AI, human cloning. There are just so many different things that can be brought to bear on this. So I can't tell you one specific person that I think is needed here. There's actually a whole range of expertise that could be relevant and different expertise is going to be needed at different times and for different things as we go forward. There's definitely a need for more scientists to be thinking about some of the open questions here, and there's definitely a need for people with expertise in policy to start thinking about how are we going to move forward.
- [125:11]
B
I can imagine some people listening, feeling uncertain about whether they should spend their careers working on other issues. So probably AI safety and governance being a big one, but also other biosecurity issues. How should those people think about whether their comparative advantage is mere biology or. Yeah, or other issues?
- [125:37]
A
I think at the moment the marginal impact of an extra person working on Mirror life is huge. If you're listening in a lot of countries, you could probably become the expert in policy around mirror life in your country within a few weeks or months of working on this. So I think I'm very happy that lots of people are working on AI safety and certainly think many people should choose that. But I would encourage people to, to think about what could they uniquely offer to this problem. I think if you're a biologist, biochemist, something like that, there's a lot that you can do on Mirror Life. We've got an expression of interest up on our website, if you're interested, please do go there and check it out. I also think if you're a researcher or a scientist who is interested in contributing to this, there are a lot of open questions that you could help with and would really encourage you to reach out. People in the Mirror Biology Dialogues Fund team have been thinking a lot about the types of research questions that would be great to have extra people working on and we'd love to help people get set up there. If you're a funder, I think there's a massive opportunity here for very cost effective funding and would love to hear from people that are interested in donating. More generally, if you're thinking that these Risks sound compelling, but are not sure how to slot in. I think building generally robust biodefences that would work against a range of threats is another thing that could be valuable to do.
- [127:05]
B
Okay, great. Yeah, we will link to that expression of interest. We've got time for one last question. So I'm interested in kind of what the discovery of mirror life and the risks it poses has kind of taught you about technological development in general, if that makes sense.
- [127:25]
A
Yeah, for sure. I think in a way this is an example, or could be an example of one of Bostrom's black balls. It's like a technology that we could draw from the urn. That could be really bad. I think I'm an optimist in general about technological development. I've worked in biotech, I've worked on vaccine development, and I think it's really important that we push ahead as quickly as possible on a whole bunch of different areas. But I do think the existence of something like mirror life means we have to be careful about some parts of the technology tree and kind of shows that they're not all going to be beneficial. So for me, it suggests we need to be humble about the types of things that might come up. This wasn't on my radar at all until 18 months ago or something, and now I think it could dominate a large fraction of biological risk. There might be other things like this that come up in the future. And so I think we need to be thinking about this, be on the lookout for them and think seriously and carefully about the types of new technologies that we want to develop.
- [128:26]
B
Nice. Yeah. Yeah. I feel really compelled by that. I think I would have. I think it is kind of one of the first examples for me ever of like, properly new, surprising technology that has this massive, massive risk profile that was just really on no one's radar except for one or two people that mentioned it a century ago. And that just feels like it should make us really, really. Well, I guess you've already used the word humble. Yeah, There could be more. And we should take weird ideas about risky things seriously.
- [129:12]
A
I am very optimistic, though, in this case that if we do the right things, we have top people working on this, we will be able to solve this problem. Another thing I think it shows is that spotting something early gives you time to think about it and to take the right action. I think this is far from a solved problem at the moment, but we can definitely solve it if we do the right things.
- [129:34]
B
Nice. Let's end there. My guest today has been James Smith. Thanks so much for coming on.
- [129:39]
A
Thanks so much for having me.