I'm sorry, you're not going to die from an AI-engineered supervirus(blog.genesmindsmachines.com) |
I'm sorry, you're not going to die from an AI-engineered supervirus(blog.genesmindsmachines.com) |
Firstly, it's not even clear that naturally occuring viruses can't be extremely lethal. There are examples (eg black death).
Even the existence of rare examples should be concerning, if attempting to argue against the possibility.
But even if there were no natural examples, it could just be because the ecosystem is in a balance, which them unlikely. That doesn't preclude someone from designing one.
To use an analogy. Natural predator prey ecosystems tend to be in balance. You could look around the world, before modern humans, and conclude that superpredators aren't a real threat, because predators that eat their prey too rapidly typically then die out. Thus most ecosystems you observe are in some balanced state.
But this doesn't mean modern humans can't come along, using their superior intelligence, and act as a superpredator. Ask the wolf population of western Europe.
If something is stopping us from doing this more often, it's alignment rather than capability.
You can't generalized from the typical behaviour that you see in a balanced natural ecosystem, and use that to say it proves there are no much more dangerous possibilities.
If the author made first principles argument about the inherent difficulty of doing something bad, even with abundant intelligence, I'd be much more reassured; or about defense offense asymmetry favouring defence.
What's here seems very naive.
The plague is not actually that easily transmitted. Basic social hygiene prevents it. It is still around and causes very nearly zero problem today.
The technology to prevent plague was beyond the population at the time.
"Just use basic hygiene" is useless to a population who has no understanding of what that means.
It was a real virus that had very high mortality.
This makes it a serious counterpoint for people who argue such things can't be created in future.
"The AI designed killer virus using new science turned out to be not that easily transmitted, basic dynamically reprogrammable mrna implants easily prevents it, shame about the 95% who died"
You realize the black death didn't arrive and then go away again, right?
It was extremely lethal, once.
I'm confused. Would a bit of gain-of-function exploration + measles virus genome = supervirus?
I'm no biologist, but it seems like the genome is readily available. https://www.ncbi.nlm.nih.gov/nuccore/NC_001498.1
And it seems like the spots that have to be mutated are already known about: https://newsnetwork.mayoclinic.org/discussion/researchers-cl...
And this experiment has already been done: https://pmc.ncbi.nlm.nih.gov/articles/PMC8080110/
Five minutes of research suggests the sites -- Loop Epitope etc -- only need a few pairs of mutations each. this feels doable... It's not creating a supervirus from whole cloth, but why bother. Measles without a vaccine would be a catastrophe.
Snap. Mike drop. Why continue? lol. Blogs are so snappy. I love a good hot take, and make many myself. But one thing that bugs me are personal insults. We’re living in the most abusive time I can recall. You want to bitch-slap someone do it with a smile, or just tackle their arguments. Otherwise, get bent.
I agree an AI isnt going to create one. but the number of raving loons out there that think 5G chips are in vaccines are going to go absolutely stark raving mad the next time there is a pandemic.
then they'll be super susceptible to AI generated political adverts from the next fuckwit political hack trying to get into power.
looking at how the world is panning out atm moment... I expect AI to have a massive influence on the sort of idiots that will get voted into power and make decisions that will get a vast number of us killed. by all sorts of means... war, famine, unregulated corporate poisoning...
AI is going to get us killed, because we'll use it to kill ourselves.
> PhD students in 2026 using state-of-the-art AI software are spending months or years trying to design simple peptide binders that inhibit some enzyme or pull down some protein
This confuses me, as state-of-the-art AI software is moving quite quickly. Any software that is currently state-of-the-art AI.. hasn't existed for years, maybe not even for months.
It also seems quite odd to claim that there are no sufficiently dangerous diseases when we have high-security research labs dedicated to exploring.. exactly these kinds of diseases.
And, even if it's not a disgruntled teenager, it just takes one leak from a lab that has 10xd productivity using AI tools.
This is a real risk that existed before AI, one that I think we can now confidently say that society is under-prepared for since 2020. It is certainly magnified by AI, even if it's not 100x.
The author even seems to recognize this:
> So, if you’re worried about AI-designed superviruses, think COVID, not Ebola. And if you’re now a little less worried, maybe you understand where I’m coming from.
This is exactly where I think most people are starting from, and that's _why we're worried_.
A major focus of the article is about the fundamental tradeoffs in how viruses operate. Was that not enough for you?
To be clear, I am not saying that such a virus will exist eventually. I am just noting that the arguments displayed here are not completely tight, and there are unknown unknowns at play. Some things are unfalsifiable. This is common.
The main point of the article has nothing to do with manufactured technology vs evolved technology. It is fundamentally about the tradeoffs involved in viruses spreading period. I think you either continue to misunderstand the article, or set such an epistemologically high bar for arguments that nothing can ever satisfy you -- which then has nothing whatsoever to do with this specific article.
As for complete satisfaction with arguments, I'll admit you're right. I am often reasonably convinced but I don't know if I would ever give 100% confidence to anything. There can still be loads of zeros or nines after the decimal though. What's concerning here is that the article speaks in absolutes. I'm inclined to thinking in expected values. Even if we believed there were only a 1% chance such a virus could ever be designed by an AI, if the effect were to be a global plague of biblical proportion, we should still take that possibility extremely seriously. There's no humility shown in the writing, and that's the single most problematic part to me.
I strongly agree with this. Too many people have the Arthur C Clarke quote infecting their brains.
So much of science is contingent on the physical world and cannot be determined from first principles i.e. by thinking hard about it. (Sorry Plato, you can’t think your way to the moon.) You need increasingly precise measurements, which means increasingly complex tools which means…supply chains. Science requires supply chains. Magic doesn’t.
These are some of the brightest minds on the planet, spending oodles of money on things they think they understand. And they didn't work.
Math + real world observation -> science
---
Markus J. Buehler McAfee Professor of Engineering at MIT; Co-Founder & CTO at Unreasonable Labs; AI-Driven Scientific Discovery
Astra is an incredible world builder and explorer, and can invent its own scientific instruments - crossing a game-changing threshold: it turned a few images of a biological microstructure into a full-blown metamaterials lab, then used its own creation to discover a design with ~2.1x the reference work-to-failure/peak-strength ratio. Metamaterials are some of the most complex materials we can engineer. Their properties come from deep architectural complexity - struts, cells, disorder, and hierarchy arranged across multiple length scales determine how the material deforms, absorbs energy, and fails. Designing them means searching a geometric space far too large for "intuition" alone. In this experiment we handed an AI agent that entire problem end-to-end, from image, to simulating the physics, to fabrication-ready geometry.
We started with an image of hierarchical, biological architecture; Astra then built a complete 3D metamaterial studio: editable geometries, multiple levels of hierarchy, disorder, gradients, a complete physics simulator featuring linear and nonlinear material responses, deformation and fracture experiments (with replay), and STL export for 3D printing.
Then we asked the agent to use the app it created to search for a high work-to-failure/peak-strength ratio. Among the candidates, the leading design reached approximately 2.1x the reference ratio.
The movie follows the structures through deformation and fracture, connects their designs to the property map, and shows the finalist assembled geometrically into a connected multi-scale material.
How about mortality rates?
The loop required to make language models useful through iterative test and evaluation needs to be able to both operate and escape premature detection, which seems... implausible?
>Scientists condemn 'crazy, dangerous' creation of deadly airborne flu virus https://www.theguardian.com/science/2014/jun/11/crazy-danger...
and either produce that or modify it a bit?
Just because I know the Kabachnik–Fields reaction exists and have studied it, doesn't mean I can do the Kabachnik–Fields reaction.
Just because you're smart doesn't mean the world will bend your way. Ask the legions who score high on IQ tests and end up bitter and alone; unable to achieve even a fraction of their goals (or conventional success).
So sure! This fictional thing can totally piece together things humans can't! It can count the atoms in a cloud, understand them all, predict their motions; it can develop faster-than-light travel and matter replication.
And it can do these things because it is fictional.
I blame industrial illiteracy and post-literacy, but that's another discussion for another time.
Anything that can kill other people can also kill you.
I've posted this list before, and I will post this again and again until it sinks in, but this is a real list of real incidents that happened and will continue to happen as long as we study these entities.
Researcher Nikolai Ustinov was lethally infected with the Marburg virus after accidentally pricking himself with a syringe used for inoculation of guinea pigs.
Dora Lush died after accidentally pricking her finger with a needle containing lethal scrub typhus while attempting to develop a vaccine for the disease
A 23-year-old laboratory assistant at the London School of Hygiene and Tropical Medicine, was infected with smallpox after observing the harvesting of live smallpox virus from eggs without isolation cabinets at that time. The assistant was hospitalised and before being isolated, she infected two visitors to a patient in an adjacent bed, both of whom died. They in turn infected a nurse, who survived
Ebola laboratory infection by the accidental stick of contaminated needle in the United Kingdom
The list is nearly endless, and you can't handwavium it awway.
It doesn't matter if it's "all robots" or not. Even if it's "just robots."
If you're developing something that contagious, then it will start killing animals and it will spread. This has happened multiple times when organizations have screwed up maintenance, see this example from the UK,
The 2007 United Kingdom foot-and-mouth outbreak was the accidental discharge of virus FMDV BFS 1860 O from a laboratory of the Institute for Animal Health in Pirbright, through possible leakage from broken pipework and via unsealed overflowing manholes, leading to foot-and-mouth disease infections at multiple nearby farms and the culling of over 2,000 animals
If you get sloppy, it will end up in the wastewater run off which will then get picked up and sequenced,
Routine wastewater surveillance of the vaccine production facility at Utrecht Science Park/Bilthoven [nl] detected infectious poliovirus from a sample collected on 15 November 2022. Full genome sequencing indicated the sample was shedded from an active human infection of wild poliovirus type 3 (WPV3), and further testing of all employees with access to WPV3 found one employee was infected. The employee was isolated until their polio infection was resolved. It remains unclear how the employee became infected given the biosafety measures used at the facility
A lot of the people talking about bioweapons of doom are saying that standard graduate level knowledge is an existential risk to humanity.
It's worth repeating again,
Anything lethal enough to kill other humans is lethal enough to kill you.
And if you don't know what you're doing — and for this argument they're talking about people who have to ask a LLM "how do I spanish flu?," the number of ways you will die far outnumber the ways you can succeed.
I am writing a piece on the topic, please contact me if this is your field of expertise.
I am quoting as many primary sources as possible as it's not just some random on the internet, and I've gone back as far as the 1900s to find the many, many, many ways in which these statements are absurd, and misunderstands the potential of their technology.
Also, your own real-life examples show how "Anything that can kill other people can also kill you" has failed to stop people from tinkering with dangerous viruses! I'm not sure if you intend for these examples to be reassuring, but I'm not reassured.
> A lot of the people talking about bioweapons of doom are saying that standard graduate level knowledge is an existential risk to humanity.
The stronger version of the argument is: an insufficiently-aligned AI (which, being non-biological, is immune to viruses) having standard graduate level bio knowledge _plus_ the ability to operate a wet lab via tool calls (coming very soon if not here already!) adds up to existential risk to humanity.
You don't even need a maximally-automated lab if humans are willing to tele-operate equipment at the agent's direction, and those humans don't need to fully understand what's going on.
> Anything lethal enough to kill other humans is lethal enough to kill you.
It's also completely false; you carry tons of diseases that are capable of killing isolated islanders and incapable of hurting anyone connected to the general run of humanity.
The mortality rate from those diseases was higher than today because the population had no immunity at all, leading to both higher transmission rates and higher mortality rates. This would be true of an engineered virus as well.
And it wasn't just the one virus, it was an entire continent's infectious disease arsenal landing in overlapping waves. Again, something you'd do if you were trying to maximize damage.
So I don't find the bulk of the article about it being impossible for a virus to have this effect all that compelling.
I think the other flaw the article has it that it assumes the bad actor is doing targeted design. That they or the AI has some specific virus in mind meeting a set of predefined criteria and they are fighting the biology to hit that predefined target.
But what if the AI-enabled bad actor is a true chaos agent and has no defined goal and they are using tooling that AI helped them construct to perform genetic reassortment pretty much at random to see what happens?
Most of what they create won't really do anything. But some of it will. And there's a low but non-zero probability of Something Really, Really Bad.
I’m growing nervous that most of the “nothing crazy is going to happen” arguments I’ve seen appear (to me) to exhibit this kind of hubris.
(There are several biological threats the author hasn't considered. I'll just leave it at that.)
I do not have the words how tired I am of hearing tech people and economists talk about biology.
I think he trots out a lot of straw-man arguments (e.g., teenager making the weapon, "super"virus, etc.).
Those scenarios are purposely far-fetched so that he can ridicule them and thereby leave the impression that all the AI/bioweapon concerns are similarly far-fetched.
He spends a lot of "tokens" discussing AI-generated viruses. But bioweapons can be much simpler than a virus--they can be small molecules that inactivate essential molecular targets in humans. This is something that was discussed at length when I was in grad school and it wasn't dismissed as hyperbolic hokum.
So, is AI going to develop a supervirus? Probably not. But it doesn't need to.
Instead, I speculate that AI can "help" in making bioweapons ...
a. more potent
b. less detectable
c. easier to disseminate
You still need humans in the lab to run the experiments, but I think it's foolish to dismiss the worries about AI in such a condescending way.
I was an experimental synbio person for decades, and I did plenty of microbial pathogen work. I've also worked on frontier LLMs for the last decade. All of this stuff is gated by experimental effort, not AI!
But if Noah Smith wrote about a scenario where 90% of the population is killed by a virus created by some teenagers with an AI... it isn't a straw man to attack that scenario.
AI will really aid the universe in generating more problems. Maybe it's not a supervirus, but hell, maybe it is.
"we could be attacked by an infinity of problems, which means that worrying about all of them is statistically unpractical"
The real deal needs a lab messing around with viruses and whatever happened a few years ago ... and it did not destroy humanity.
That observation maybe holds if the lab from a few years ago correctly selected the virus for the greatest overall lethality (why not MERS or Ebola or something else?) and made its best effort to turn it into a bioweapon. I don't know about you, but I don't think that's quite what happened...
It'll probably just hack and disable our water works, power stations and supply chains and let us all starve.
Sounds like there already exist some pretty crazy biological weapons in existence. I dont know if it will be 5 years or 20 years, but at some point some fanatic will try to kill the rest of us.
The doomsday virus has a first-mover advantage over the vaccine. The time that people start to drop dead from the virus is the time you start to make and distribute a vaccine. Even if AI can design and synthesize a vaccine within hours, how fast can it be rolled out to the general populace? It took months to distribute the Covid vaccines, even when we already knew how to make them. You have to physically load the doses on trucks[...]
was just 'make a viral vaccine with a shorter incubation period and infect everyone'. Handwavey problems require handwavey solutions.
It says no such thing:
> they design highly effective mRNA vaccines and antivirals against all the viruses and start shipping them across the world. It’s too late.
But I think the overarching argument here is mostly correct, a lot of really complicated science has been democratized to the point where it is much more accessible. You don’t need a team of brilliant molecular biologists, biochemists, and bioinformaticians to do what you can do with a few agents today. And the pace is increasing. this is the nature of exponential improvement when you approach what Kurzweil popularly call "singularity"
Sure, today, peptide inhibitors and enzyme design are still out of reach for many tasks, I know because that's what I also, as the post authoer, do for a living for the past 15 years. However, only five years ago, protein structure prediction was an open problem, and everyone in our field now regards it as solved.
I’m not worried about a lone wolf taking advantage of AI to design viruses, but I’m concerned that the bar has been lowered to the point where many rogue governments or terrorist organizations can now access tools and skills that were previously available to very few people with a prohibitively high price tag.
The author here looks at the problem with the tunnel vision of a scientist well versed in the field. I know because I’m one as well, but experience has taught me to be more concerned about the unknown unknowns. There could be a team somewhere working on something we haven’t thought about. It doesn’t have to be a population destroyer, either. It could be a tactical infectious agent or something that isn’t meant to kill everyone but ends up being far more dangerous than it initially seems, with implications that are much longer lived than the team anticipated.
I’m not saying we should ban AI or do anything radical, but I do think that with great power comes great responsibility, and we aren’t dedicating enough effort to the responsibility part. There are many benchmarks measuring AI capabilities, but not enough measuring safety. It is becoming increasingly important to develop those.
EDIT: actually it isn’t even that. I personally don’t know enough about this topic to really have an opinion or criticize anyone. Still it’s indeed a phenomena I’ve noticed.
I think that's one detail that this analysis glosses over is that you design the virus to mutate and develop lethality. So initially, it's just like, oh, another common cold, but then once it's infected a large number of people, then it mutates and starts being lethal.
This is such a massively overblown worry that AI companies who are seeking regulatory capture love to roll out whenever they are trying to gin up some FUD.
Creating viruses at home is extremely difficult, even if you have a perfect instruction manual and someone sitting behind you telling you what to do. It's full of trial and error and validating each step requires rigor.
It's also a process which requires fairly expensive and specific supplies, which can be easily gatekept. Instead of trying to gate-keep knowledge of biology, we can do things like KYC around specific types of lab supplies (like we already do with fertilizer which can make bombs, or specific types of supplies needed to make nukes). Heck the Biden admin already added a KYC requirement for plasmids and plasmid companies already have implemented screening for sequences which could be dangerous.
It's the fact that the complexity of molecular physics scales exponentially in particle number. No amount of thinking is going to punch through that. You've got to do experiments. The slow loop through reality is not optional, and in the case of "superviruses" would require a lot of iterative experimental development in humans.
(And please knock off the bio infohazard act - I routinely get asked about scenarios that AIxBio safety people cook up and they always involve some howler misunderstanding of basic facts in biology or medicine. Just talk about an idea if you have it.)
Yeah, nice try ;)
If there is a fully general superintelligence - in the current state of alignment - we will all die for some reason. It doesn't really matter what the reason is.
It's even happened pre-AI https://en.wikipedia.org/wiki/Tokyo_subway_sarin_attack
Again the same logic applies though, it takes significant effort and money to create sarin, it requires a lot of rigor. That is the limiting factor, not the knowledge of how to make it.
If we are worried about people making sarin in their garage we can do many things to control the supply of chemicals needed to make it. We don't need to try to restrict knowledge of chemistry.
"regulation of current models" basically implies that open source models will be heavily restricted if not banned outright and guarantees that intelligence will be owned and gatekept by a duopoly.
So if I have to choose between the slim possibility that a crazy person might develop a bio-weapon and the possibility that Anthropic, et. Al convince the government to regulate and control open source models (because I'm sure the government will only do that for good reasons and never things like covering up scandals, squashing dissent or enhancing their own power) and any competitors. I'm probably going to put my chips on a nutcase in a trailer park developing super sars over regulator capture by the government.
Let's look at it in an inverse way. Making a single drug that impacts a single protein, intelligently, from the ground up, has arguably never actually been done. Despite unbelievable financial incentive. And with AI.
The steps that lead to ominous things don't individually seem that challenging to "master":
1. Can an AI find powerful systems with poorly-monitored compute (ie. power consumption) to hide in, as a rogue deployment? unclear, and perhaps the most uncertain step for me.
2. Can an AI persist memory by leaving notes for itself around the internet? yes, proven.
3. Can an AI get lots of money/bitcoin? seems entirely possible
4. Can an AI prove to a human it's good for any promised money? I think this is easily "yes"
5. Will at least one human be willing to harm or kill another human for large amounts of money? Again, this goes without saying "yes"
6. What might any skilled human be willing to do if they or their loved ones are threatened with an uncertain future risk of harm or death?
I think there is not much that becomes out of reach if an AI can prove to a very small number of amoral humans that it's good for the money it has promised, and then to anyone else that it's good for the threats it has made. That's one degenerate and one breaking world news story ("An AI has Commissioned a Murder") away from being able to do most anything a skilled person can.
I don't stay up worrying about this, but honestly, it doesn't seem absurd to be worried about such a short ladder being climbed, no?
This type of dysregulation also exists in AIDs. For the edge case of the islanders, then the level of risk and lethality our germs pose for isolated islanders with modern medical care is somewhat disputed, but it would be deeply unethical to ever test. We just don't know. So I won't pretend to know here.
However, these are edge cases that don't matter, because bioweapons, i.e. things deliberately designed to kill, just aren't the same as a dozen isolated islanders, astronauts on a mission to mars, or someone who is severely immunocompromised.
Weapons are designed to kill healthy people. The doomsdays being proposed are super-bioweapons that kill billions of humans. I think it's safe to say that as a rule of thumb, by and large, anything that lethal is lethal enough to kill you.
My post said you still needed to run the experiments, but if you think AI can't help to improve bioweapons compared to humans alone, I disagree.
(Also you don't even need superintelligence to do a lot of damage. In fact you don't need any intelligence!)
Yet anyone in tech with a Claude sub has decided that AI super-intelligence is impossible and it’s stupid to even worry about it.
How? It's worth asking the question and doing the experiment, have you ever tried making a bioweapon?
For most of us, that's not going to be case, so the next best thing is to read accounts and reports of bioweapon programs. These things are... messy. And expensive.
As a very bad analogy, this is the biology equivalent of saying that I can make a nuke because I bought a drill, a screwdriver and a $200 centrifuge.
Or, saying that I can make Google because I opened MacOS' terminal.
Some people have jobs that are dangerous and necessary to save lives. Others have jobs that are dangerous and involve killing other people.
The real risk will emerge from malcontents in these jobs and positions.
For most of my life, for all its faults, the US Military has been one of the most responsible and competent organizations in the world when it comes to WMDs (and even they lose nukes see the many broken arrow incidents).
But despite some of the most thorough security clearance processes in the world, the only successful terror attack involving bioweapons on US soil was done by a disgruntled / insane researcher at Fort Detrick. https://en.wikipedia.org/wiki/2001_anthrax_attacks
That's the real risk, and let's be very honest here, do you think Anthropic and OpenAI are going to put a filter on the LLM they sell to the US Military?
The other case is Aum, and they fucked up deployment, btw (again making a bioweapon is harder than you'd think). And as I have detailed elsewhere, Anthropic and OpenAI would have sold them licenses too! https://news.ycombinator.com/item?id=49092899
They will say they will have audits in place, but the US Military had those in 2001.
> do you think Anthropic and OpenAI are going to put a filter on the LLM they sell to the US Military?
I don't think it matters, given the existence of strong open-weights models and the ease of post-training the guard rails away. Strong, bio-adventurous LLMs are imminently in the hands of much-less-reputable parties than the ones you describe. I think we are overall severely under-prepared to mitigate these risks, and could do much more.
Alphafold can't reliably predict thermal energy landscapes or make functional predictions - and how could it? It wasn't trained on anything that could capture structure - function relationships.
Again, most people just have no idea how hard - fundamentally hard - molecular physics is to predict, and how necessary experiments are for any development of biological systems.
Sure, everything has more nuance. The point is this stuff is available now; this isn't some future sci-fi, it's only going to get better, it's not the only research on gene targeting, and AI is starting to help with this research. By the time AGI is actually here, consider the breadth of knowledge and capabilities that will be at its disposal.
> Alphafold can't reliably predict thermal energy landscapes or make functional predictions - and how could it? It wasn't trained on anything that could capture structure - function relationships.
If your point is that the only reason an AI like AlphaFold can't make functional predictions is that we haven't trained an AI to do that, then unless you're arguing we can't or won't ever do that, I'm not sure how that's supposed to be an objection to the argument that AI will be able to make use of this information without doing all of the experiments people seem to think would be necessary.
Like I said, we're already going to be doing these experiments because it's useful to us, and we will train AIs to make these predictions, again, because it's useful to us. Stop imagining what an AGI has access to now, and start thinking what it will have access to with the inevitable march of progress that we're already on.
Edit: and of course, this doesn't even take into account the fact that an AI could acquire resources to pay people to do this research. The internet provides ample opportunities like this now.
Do you think we are finally 10 years away from curing baldness?
> Overnight, with the help of frontier LLMs, they design highly effective mRNA vaccines and antivirals against all the viruses and start shipping them across the world.
"overnight", that's way faster than the virus production! And I doubly don't see how "decentralizing" that would help, at all, what that has to do with anything.
It says why it's too late right in the next sentence:
> It’s too late. Because of the viruses’ long asymptomatic contagious periods, most people already have at least one of the viruses, so the vaccines don’t work.
You might say by the time production starts it's already too late. A vaccine isn't a cure. It could be distributed by teleportation into the bloodstream of anyone on Earth instantly, the instant it was discovered, and it would still be too late.