Yves here. We published a GPENewsdoc piece on AI synthetic biology which highlighted key risks. Sadly the post got very comments, perhaps because readers found the discussion to be too academically oriented. This piece is much more layperson accessible, although some may dismiss it as simplistic. But having seen more modest biomedical advances like CRISPR and stem cell therapies sometimes produce bad outcomes, caution is warranted.
By Kurt Cobb, a freelance writer and communications consultant who writes frequently about energy and environment. His work has also appeared in The Christian Science Monitor, Resilience, Le Monde Diplomatique, TalkMarkets, Investing.com, Business Insider and many other publications. Originally published at OilPrice
- The combination of AI and synthetic biology could lead to the creation of novel organisms with unintended and potentially harmful consequences.
- The democratization of genetic engineering through AI-powered tools raises concerns about bioterrorism and the accidental release of dangerous organisms.
- The use of genetic data for AI training raises ethical concerns about biopiracy and the exploitation of genetic resources without consent.
Science fiction films are replete with human space travelers visiting far-away planets that have atmospheres suitable for those humans to breath. Thus, the bother of wearing a space suit or other protective gear is dispensed with, and the encounters with alien races, both hostile and friendly, can proceed without such cumbersome gear mucking up the works.
In addition, these planets often have plants and animals that are strikingly similar to those found on Earth. The problem with this all-too-frequent occurrence in science fiction stories is that even if such planets exist, they would have microorganisms entirely unfamiliar to the human body and thus likely to kill it within days or weeks. Humans would have no immunity and suffer a fate similar to that of the indigenous people of North and South American when Europeans arrived bearing diseases unfamiliar to indigenous immune systems and therefore profoundly deadly. Up to 90 percent of the natives perished.
Enter synthetic biology, that is, the engineering of organisms never before seen on Earth. We’ve already seen it in the form of genetically engineered crops such as GMO soybeans and corn. But that is a pale forerunner of what is about to happen: the marriage of artificial intelligence (AI) and synthetic biology. For many years already scientists have been able to create novel sequences of DNA, and they’ve already created dangerous designer viruses for research purposes. I’ve written previously about the possibility of systemic ruin that can flow from these activities. And I’ve voiced concerns about the democratization of genetic engineering through do-it-yourself kits: “Anyone with a credit card and a mailing address can now order their own genetic engineering kit.”
The emerging democratization of genetic engineering will be supercharged as AI is married with synthetic biology. Imagine being able to type in normal English (or your native language) a description of the type of synthetic biological organism or substance that you want and have it produced on the spot. This is not yet possible today, but there is no reason to believe it won’t be in the years to come.
One of the critical problems is described in the previously linked article above:
Large language models such as ChatGPT routinely incorporate elements in their output that appear compelling but are factually inaccurate or bizarre: Living people are described as deceased, dates are given wrongly, generative AI images of people develop additional body parts or mangled unreadable signage is created etc. While such hallucinations and black box failures can be problematic enough in the 2 dimensional and electronic domains of text, image, video or sound, they could be highly problematic if incorporated into genomic design of four dimensional living organisms or of active biological proteins released into the body or the biosphere.
I have hammered home again and again in my previous writing that what is called “the benign assumption” prevents most people from recognizing the dangers of new technologies. They simply believe that new technologies will be used with the best of intentions, and, of course, the seeming positive outcomes of such innovations are heavily touted in the media by those who stand to profit handsomely from the unbridled dispersion of the new technology and its products.
Trouble is, this is the same thinking that has brought us the widespread crisis of toxic chemicals everywhere in our food, water, air and soil. It’s the same thinking that brought us the adoption of social media by nearly everyone with an internet connection in the world. How nice it will be to easily and quickly share our lives with friends and family and make new friends across the globe! How nice it will be to share useful advice from every walk of life instantaneously with millions and even billions at virtually no cost! And, those things happened! Then came the dark side: cellphone addiction, the spread of organized hate and misinformation, the targeting of individuals and institutions for harassment, and the killing of people by mobs because of false accusations of “child trafficking, organ harvesting or other egregious acts.”
The dark side of AI plus synthetic biology is already visible. One focus of synthetic biology is the creation of novel proteins that might be used to treat disease or substitute for proteins we currently get from both plants and animals. But it’s also possible to create toxic proteins or ones that are harmful for other reasons, intentionally or by mistake as described above.
To the extent that novel proteins become additives in our food, they tend to reinforce our dependence on processed foods that are now being widely recognized as a major cause of chronic disease. By definition such proteins cannot be whole foods that our bodies are adapted to thrive on.
These novel substances can also be used to produce substitute foods. One major area of application is allulose, a sugar found in figs, raisins, wheat, maple syrup and molasses that is not absorbed by the body. But it’s found is such small quantities that it remains expensive. The synthetic biology industry is working to create enzymes that will turn cheap feedstocks like starches and sugar itself into allulose. But, this hardly seems like the healthy eating breakthrough we need to tackle the chronic disease epidemic.
Beyond the health effects and the possibility of systemic ruin because of bad actors, there is the issue of biopiracy. The information on genomes of hundreds of millions of samples of genetic material used to train AI systems comes from around the world. No one has given their permission for these genetic resources to be used by the AI industry exclusively for its own benefit. It’s similar to the mass ripoff of the works of artists, writers, musicians, journalists, software designers and other creative individuals and organizations by the current AI purveyors who train their text and image generating systems using the work of millions of people without permission. (By the way, the AI industry is being sued for this practice. See here, here and here.)
It should not surprise readers that AI depends on what is essentially open theft of other people’s work. So, it won’t surprise you that the same strategy is being deployed in AI’s marriage with synthetic biology. But there is a big difference between the two areas of exploitation. The first threatens the livelihoods of all creative people whose work is available on the internet (and possibly in easily scannable materials). The uncontrolled dispersion and application of the second threatens us with mass death and even extinction.
The titans of the AI industry and their naive or cynical supporters will tell us that in order to bring the as yet unknown countless benefits of AI combined with synthetic biology to society, governments need to stay out of the way. But anyone with a sense of history knows we’ve been here before countless times. Asbestos, leaded gasoline, chlorofluorocarbons, and Teflon come to mind. Anyone with a sense of history knows better.
If you go back in time, AI was a hot idea in the decades immediately after WWII. Much of the iconic sci-fi written and filmed imagining a future full of intelligent robots dates back to that time; what was created after that never really had the same impact.
However, the technology turned out to be much behind the curve imagined back then, so the subject kind of faded way from collective consciousness. Together with the collapse of the space programs in the 1970s, perhaps.
But now it has reemerged as a practical reality. And here lies the problem.
First, things like synthetic biology weren’t on anybody’s radar back then, so there was no Asimov to write any three laws on it.
Second, and much more important, pretty much all of that foundational thinking from way back in the days implicitly assumes some kind of communist, post-scarcity society in which these things will exist. Which automatically reduces a lot of the dangers associated with them. Because those post-war decades were a time of unprecedented economic redistribution, even in the capitalist West.
What we got instead is AI actually emerging in a hypercapitalist society (after half a century of vicious neoliberalism) that is just about to start sliding down the decline part of the resource availability curve, with several decades separating us from the last time society seriously thought about these issues, and with zero control over what is happening, it all being in private hands.
About the worst possible situation imaginable…
“A new species would bless me as its creator and source; many happy and excellent natures would owe their being to me. No father could claim the gratitude of his child so completely as I should deserve theirs. ” – Victor Frankenstein
>Anyone with a sense of history knows better
But “They” won’t because no common sense either. I hate to be negative but the simple question: “What could go wrong?”, seems to take a back seat to: ” How much will this add to our quarterly earnings?”
Given the current state of the world, I wonder if perhaps homo sapiens is an AI hallucination.
When you hear the words “the democratization of…” you know you are the mark.
I think this is example of omnipotence fallacy. The classic illustration is question if God can create rock heavy enough that he can not lift it himself. If he can create such rock, then he clearly isn’t omnipotent, because there is something beyond his abilities, i.e. to lift it. On the other hand if he can lift everything, then he is limited in his abilities to create things, which also leads to the conclusion he isn’t omnipotent. In other words the idea of omnipotence is inherently contradictory. Now the standard religious retort is that God is not bound by logic and doesn’t care about thought experiments of puny humans, but for AI there isn’t such escape hatch.
So if the proposition is that it will be possible for random kid to type into Google he wants virus that can kill all of humanity and in 20 min the vial is at his doorstep, then it will be possible for another kid to ask for virus/bacteria that can neutralize all viruses able to kill humans, and get it in 15 min, because she is closer to Amazon fulfillment center.
Nature routinely mixes genes and, very very rarely, comes up with entirely new genes. But nature has its limits, and there are a lot of things that are simple extremely improbable to occur by natural selection.
But creating new genes/proteins/organisms – even without AI – is another story entirely. While we will likely never know for sure, I would lay odds that COVID was created in a lab. I mean, the lab in Wuhan had proposed to create just such viruses (and almost certainly was), COVID started in the one city where such viruses were being made, I mean, seriously. And we may never be rid of COVID.
Now we are going to do this on an industrial scale?
I note that in basic science gene modification in mice are a regular tool. But of course now and then mice will get mislabelled or put int he wrong cage etc. It is a dirty secret of modern science that the laboratory mouse if a hodge podge of past genetic experiments. So if we have (say) does that are designed to secrete industrial chemicals, and these inevitably get into standard dairy herds…
The work should be banned outright. There is no other way to control it.
The problem is that the profit motive leads to recklessness.
The solution is to remove any AI/Synthetic Biology products from IP protections, which removes the profit motive.
Thanks for the synbio coverage Yves. I didn’t comment on the last one because I had nothing of value to add.
It will be interesting to observe how this technology plays out while the social contract continues to disintegrate and access to energy likely becomes prohibitively expensive to most.
The “dark” side of AI-powered synthetic biology …
I wonder what the “bright” side of it might even be? I haven’t heard of it yet.
Humankind has played with genetics for millennia. The efforts of our distant ancestors created most of the food sources we enjoy today and lead to the creation of agriculture. The efforts of Science discovered some of the design of life and invented powerful tools to use for extending and powering Humankind’s play with genetics. Neoliberal Market theology has crafted a world of humans controlled and driven to generate profits unencumbered by ethical considerations or concerns for the long-term. Our Corporations act like an unhuman Midas largely untroubled by consequences beyond next quarters profit reports. The age of fossil fuels has wrought many unintended consequences threatening Society, Civilization, perhaps the existence of Humankind — Climate Chaos, resource exhaustion, the pollution and poisoning of Earth, Air, and Water. The age of fossil fuels has also enabled Humankind to discover the powerful tools available to play with genetics, and synthetic biology. In the hands of the Corporate monsters these tools can and have too easily created additional unintended consequences. The government that might guide and control the Corporate monsters has become their wholly-owned property.
Life can and has adapted to radical changes in the climate. The changes taking place in the climate in the future will occur at a speed exceeding the speed that life can adapt. I believe the tools for extending and powering Humankind’s play with genetics are some of those tools Humankind would need in efforts to adapt and survive in the future world. Those tools are held in the wrong hands. Remedy, if there is to be any remedy, will not be gentle or small.