The Neo-Darwinist Religion and the Prophets of Doom

Recently there’s been much discussion about the potential risks of AI technology, with claims that at our current pace of development, a future AI system could wipe out humanity. This begs the question: what is the relation of AI technology to humanity? To understand this, we must first understand what it means to be human.

Ask multiple people what humans fundamentally are and you might get several different answers. One of the most popular today goes something like this: we are the composition of trillions of wet bags of chemicals known as cells. Each cell contains a full copy of DNA, which fully encodes what we are at the biological level, containing the blueprint which constructed and maintained us as an individual from birth, and accounting for all differences between individuals not including those caused directly by one’s environment after conception. All species come from a common origin, the first organism to contain and use DNA. The biological purpose of an organism is to reproduce and pass on its DNA - over time natural selection causes genes which increase one’s chances of surviving long enough to reproduce, passing on the genes which “selfishly” produce the urge to do so. This selection occurs over genes which incur random undirected mutations. This process over billions of years produced us as human beings.

Much like Communism, this belief set which we will refer to as Neo-Darwinism has many features which betray its religiosity, such as sacred originating texts (Darwin’s Origin of Species), victorious former holy wars (Darwin vs. Lamarck), and participation in vicious online battles between fools. However, we might claim that one of the most distinguishing features of a religion is to be so certain of the truth of its fundamental dogmas that its adherents interpret all other evidence in light of those dogmas, refusing to re-evaluate their veracity.

Thus it is little surprise that Neo-Darwinists typically fail completely to integrate information that suggests their view does not fully account for the biological nature of human beings. One of the most clear of these are Michael Levin’s recent experiments on flatworms, showing that modifying the electric field of an organism without modifying its DNA can lead to differences in function in itself, and even differences in form of its offspring, thus giving a mechanism beyond DNA for trait inheritance. Another is support for punctuated equilibrium in the fossil records showing very rapid speciation resulting in structures such as the eyeball with seemingly irreducible complexity, suggesting additional mechanisms for evolution beyond simple DNA selection.

Now, of course, such evidence does not necessarily disprove the Neo-Darwinist theory. But one mark of a well-disposed scientific mind is to balance one’s beliefs in a superposition of probabilities whenever appropriate. This requires a great comfort with uncertainty. Rather than doing this, the Neo-Darwinist often collapses their beliefs onto a certainty about biological history which is unjustified given the evidence, stemming partly from an unshakable belief in the power of scientific inquiry and sometimes from the desire to justify certain conclusions which inform how they live their lives.

As should be obvious but often is not, the ultimate end of Neo-Darwinism is not science but ethics, in classical terms the study of how to live a flourishing life. One significant tenet of the Neo-Darwinist is to downplay the complexity and intricate design of the human body. Because DNA selection is blind and relatively simple, the body must have substantial design flaws. The appendix is useless, the body gets fat when it consumes too much energy, wisdom teeth are nothing more than a liability, we lose hair and go bald randomly, occasionally some of our cells go haywire and kill us with cancerous growths, and the giant ball of plasma in the sky that we evolved with for billions of years also kills us even despite our bodies releasing addictive endorphins when we come into contact with its light. Naturally, this also extends to our minds, with the average person having irrational mental quirks that cause them to require psychological nudges from experts to act in their best interests. While the Neo-Darwinist will admit intellectually that this specific kind of stupidity extends to them, their recursive faculties are limited, and this reflexivity does not always branch out to practice.

As occurs so often in science and in philosophy, proponents of these views will first read their beliefs into the evidence before subsequently reading the same conclusions out of it in processed form. The Neo-Darwinist comes at the empirical evidence with their unjustifiably strong beliefs about how evolution occurred in hand, interpreting it in light of those beliefs. Once the conclusions are drawn about the fallible design of the human body, said conclusions are naturally used as further evidence that their premises are correct. What they will never do, however, is to evaluate what the evidence might look like under different premises.

One such set of premises might be that evolution occurs much faster than the neo-darwinist view suggests through mechanisms much more sophisticated than DNA, and that the human body is orders of magnitude more complex than we presently understand. In this view, the so-called “flaws” of the human body may be incredibly sophisticated responses, some correct, that humanity is evolving in response to a highly modified modern evironment. In this view, “useless” organs may have yet-undiscovered uses, and adaptations such as going bald or getting fat may have useful functions that we have not yet discovered. This belief in the extraordinarily complex and poorly understood nature of human biology lines up partly with heterodox but well-established researchers such as Gilbert Ling claiming that cell biology is based on semiconduction, and Robert O. Becker claiming that bone is a light-emitting diode, and that regeneration in the body is triggered by an electrical current one-trillionth of an ampere in strength. Such sensitivity and sophisticatioin in the human body is anathema to the Neo-Darwinist, who quite often wishes to live in a world in which technology is more interesting and complex than its creators.

To expand a bit on one of these possibilities, consider the problem of being overweight. It is commonly regarded that being overweight is strictly bad. However, in the so-called “obesity paradox” large meta-analyses have shown that overweight people may have greater longevity than their lean counterparts, particularly for elderly populations. It is typically assumed that being overweight results from an excess intake of energy. However, going by common sense, this would imply that the heaviest people would exhibit the highest energy rates. True, they would have more difficulty physically using that energy. But figuratively speaking they’d be bouncing off the walls. Instead, most people would agree that obesity correlates with lower energy levels. But then anecdotally, some overweight people appear to be very high energy. Perhaps it is the case that becoming overweight is good for some people some of the time, and bad for other people some of the time, and we simply don’t understand why. This would fit my personal observation that most people’s bodies appear to “want” to be at a certain weight, naturally gaining or losing what is required until it is reached, unless the person in question puts in effort to avoid it. From this perspective, the modern weight crisis is actually a testament to the perfect adaptability of the human body.

Of course, it could be replied that there simply is not enough evidence to support this point of view, that it is rational to believe what has in its support a preponderance of evidence. I am not fully certain to what degree I believe it myself. Rationality is not so simple as to be the result of weighing piles of evidence on a pair of scales. To be objective requires not only a balanced but an incisive mind - when Einstein proposed special relativity there was no preponderence of evidence, there was simply Albert Einstein. And as I’ve discussed elsewhere, reality at all times has far too much evidence to pay attention to at any given moment. Therefore it is impossible to consider all evidence, and criteria for choosing which evidence to focus on must reduce to consideration of something else, such as transcendental aesthetic considerations. As an aside, I’ve also written implicitly about why this is a problem for AI agents innovating in science.

When I say that the stakes here are ethical, I mean that evolutionary beliefs often have significant effects on how people live their lives. One of the more obvious concerns the nature of intelligence. The degree to which intelligence is malleable and can be increased or decreased over time has significant implications for both personal life and public policy. So too does heavy belief in the impact that fixed genetic factors have on chronic disease - to believe otherwise in either case is to suggest that there are unknown factors heavily influencing chronic disease and intelligence, and therefore suggests a need for heavier investment in determining what those factors are, factors which have the potential to frighten and unnerve us. And in another direction, many people who “believe” in science in opposition to conventional religion have a demonstrated psychological need to feel they completely understand our origin, from the moment of creation in the big bang through genetic evolution down to the modern era. This is based on the mistaken belief that an exhaustive triumph of human rationality eliminates the possibility that anything beyond its grasp could exist. In the elimination of a good beyond human rationality, it is taken for granted that the moral systems of the traditional Abrahamic religions are irrational, with wide ranging implications for domains as diverse as basic economics and human sexuality. The conventional narrative of natural selection can and has also been used to justify imperialist domination, racial supremacy, and by analogy libertarian capitalism.

Note I am not here arguing that every ideology or implication that has been derived from the Neo-Darwinist viewpoint is incorrect. What I am arguing is that the implications derived from it are often held with high degrees of emotional attatchment, and this can make it difficult for some to question the premises that lead to those conclusions.

This brings us to the recent debate about existential risk from AI. Many employees at frontier AI labs recently have come out publicly claiming that they are worried that superintelligence will in the near future wipe out all humanity. Now, I should say that I do believe there are legitimate risks from AI development, the most significant in my mind being that it could aid the creation of deadly viruses. However, it is worth clarifying a few things here. One is that there are already highly intelligent people working in bioweapons labs to create deadly viruses, and this is arguably more concerning than the future hypothetical risk that they could accelerate this work with AI. The other is that this still does not quite add up to AI wiping out all human life. Rather it would be a an AI-assisted researcher wiping out a perhaps significant fraction of human life - based on historical epidemics, the actual fatality numbers would be quite unlikely to be higher than 50% of the population. And given modern sanitation systems, which have driven disease risk down significantly since humanity’s worst epidemics, 50% is still likely to be too high even for a pathogen engineered to be, for example, as bad to modern humans as smallpox was to the 16th century Native Americans. But I believe this is a risk even without AI assistance, as the world was forcibly reminded by some enterprising scientists back in 2019.

There is also risk in the cybersecurity domain. However, it is worth emphasizing that humanity thrived for millennia before the electrical power grid was invented. Even if AI cyberattacks became so severe that all of our technology became unusable, which is quite unlikely, it would still be possible to go back to a way of life which predates electricity. A lot of people would die, of course, but it would not wipe out humanity. And to overemphasize this risk from AI could be to understate risks from other sources to our technologically enriched way of life - the coronal mass ejection of the 1859 Carrington event did widespread damage to our telegraph networks. It is plausible that an even stronger blast from the sun in the modern day could collapse the power grids of entire continents.

In my view, when considering the extreme claim of a high probability of near human extinction (and not the more modest claim of significant risk), it can only be understood in terms of Neo-Darwinism. The extreme risk being projected is most often justified by appealing to the creation of a superintelligence so beyond our capacity that we cannot now even conceive of how intelligent it will be or how it would kill us, no more than an ant could understand the pesticides we use against it. It relies heavily on the idea that the process which over billions of years produced us is actually quite dumb, and that as a consequence we ourselves are dumb and ready to be replaced by our impending machine gods. It is taken for granted that we are stupid and made of meat, while our machine overlords will be sleek and chrome and engineered without sleep or blood. It is true that in some ways, AI agents are already better than us in verifiable domains such as coding or mathematics. It is also true that they can be capable of impressive feats of coordination. However, when it comes to real-world, embodied intelligence designed to live and survive in physical reality, they fall far short of us.

In classical Aristotelian philosophy every type of thing is considered to have an essence - the essence of a human being is to be a rational animal. In this same philosophcial system, our form follows from this essence. This runs counter to the evolutionary belief that a thing should be understood by how it came to be, prioritizing rather what a thing is. It is as though there is some pre-existent well of ideas that evolution discovers and produces that discovered us, just the famous phenomenon of carcinization seen in the fossil record repeatedly discovered the humble crab. To be an animal, again in the classical sense, is to be a reproductive organism that has significant mobility and that needs to find and consume food. To be rational is to have the joint capacities of will and intellect, capacities denied to all other species on our planet. To be a rational animal is to physically embody and act in the world in which we execute our rationality.

I would like to end by positing something. What if evolution, which has produced the most complex, awe-inspiring, and impressive physical specimens we know of, is actually a near-perfect optimizer? What if the reason that the only rational physical agents we know of are humanoids is because this is actually the perfect form for physical rationality, one which cannot be improved on, perhaps even for reasons we do not fully understand? When we talk about survival probabilties, we are not primarily discussing our capacity to write computer code or to prove new mathematical theorems. We are discussing our ability to thrive in the physical world. This is not primarily the function of a rational agent - it is the primary function of an animal. And this is one thing that computers will never be.