Smaller Brains, Smarter Machines

What Evolutionary Biology Says About a Future Shaped by AI

#AIEvolution #HumanEvolution #ArtificialIntimacy #SelfDomestication #EvolutionaryBiology

Warm-Up (5 minutes): Think about the last time you let an app remember something for you, a phone number, a route, a fact you used to just know. Now imagine that habit repeated billions of times a day, across every generation from now on. Write down one mental skill you personally rely on AI or your phone to handle instead of your own brain. Keep that skill in mind. This lesson argues that small shifts like this one are not just changing how we live day to day, they may be quietly shaping which traits get passed on to future humans.

Who This Is For: This lesson is for evolutionary biologists, AI ethicists and technology policy researchers who study long-term societal impact rather than just near-term risk. It also serves product designers and executives at companies building AI companions, dating apps and social platforms who want to understand the deeper behavioral forces their products may be triggering. Sociologists, demographers and public health researchers tracking mood disorders, mating patterns and cognitive change will find a rigorous framework connecting these trends to natural selection. Anyone curious about the difference between AI's impact on your life today and its impact on humanity across generations will find a clear, scientifically grounded answer here. The shared challenge across these roles is thinking past the next product cycle toward effects that unfold over centuries.

Real-World Applications

This lesson argues that AI is already functioning as a selective force on humans in the same way domestication shaped dogs from wolves and corn from teosinte. It points to concrete, already-existing technologies, AI matchmaking apps, virtual companions and AI-assisted fertility treatment, as mechanisms actively influencing who mates with whom, who reproduces and which psychological traits confer advantages in an AI-saturated world. This is not speculative science fiction about robot uprisings, it is a biologist applying the same tools used to explain finch beak evolution and wolf domestication to dating apps and chatbots. Companies building AI companionship products, fertility clinics using AI-assisted embryo selection and criminal justice systems using AI in sentencing could nudge human evolution and exert longer-term genetic effects.

Lesson Goal

You will understand the three-part framework Brooks uses to classify how AI could influence human evolution, ranging from deliberate genetic selection to completely unintended side effects of everyday AI use. You will be able to explain why he predicts smaller human brains, changes in attention span and shifts in personality traits as plausible long-term consequences of AI adoption. You will leave with a clear sense of how AI-powered dating apps and virtual companions are already altering mating competition in ways that could shape future generations.

The Problem and Its Relevance

Most public discussion of AI and human evolution jumps straight to dramatic scenarios like extinction or a technological singularity, but this lesson argues the far more likely story is a slow accumulation of mundane, everyday nudges that operate exactly like the inadvertent selection that turned wild wolves into golden retrievers. A separate and more unsettling problem is that AI applications built for entirely different purposes, dating apps optimizing for engagement or criminal justice tools optimizing for efficiency, can end up functioning as agents of eugenics even when nobody involved intends anything of the sort. Together these two issues mean that the biggest evolutionary risks from AI may be the ones nobody is designing for, hiding inside products explicitly built to entertain, match, or assist us.

Why Does This Matter?

• Natural selection does not require intention to operate. Brooks explains that any factor influencing who lives, who mates, and who successfully raises offspring will produce evolutionary change over generations, regardless of whether that factor was designed with reproduction in mind, which means AI does not need to be built for this purpose to have this effect.

• AI dating and virtual companion technologies are already changing mating competition. If AI companions mostly attract single men, fewer men are left competing for dates, so dating gets less competitive. If they mostly attract single women, fewer women are left to date, so men compete harder for them. 

• Brain size reduction is a real historical precedent, not speculation. Human brains shrank at a rate estimated to be fifty times faster than the rate at which they grew, likely because collective knowledge and shared culture relieved individual brains of some cognitive burden, and Brooks argues AI cognitive offloading could accelerate this same relaxation of selection.

• AI in criminal justice can quietly function as eugenics without anyone intending it. Because incarceration affects who has the opportunity to find partners and raise children, and because many traits are at least partly heritable, AI sentencing and parole algorithms that inherit historical bias could shape gene frequencies over time even though no one designed them for that purpose.

• Self-domestication, the process that made humans cooperative and social, could be eroded by the same AI that mimics its benefits. Brooks argues that facial recognition and surveillance AI could make it harder for ordinary people to conspire against violent despots, a capacity he identifies as central to how humans became a cooperative species in the first place.

Core Concepts

This lesson starts from a foundational idea in evolutionary biology, that natural selection is simply what happens when some individuals in a population leave more descendants than others, and that this occurs whenever some trait consistently affects survival, mating, or reproduction. It borrows a framework from Charles Darwin's own writing on crop and animal domestication, dividing selection into three categories, selection in nature that happens with no human involvement at all, inadvertent selection where humans unintentionally influence which individuals thrive, and deliberate selection where humans consciously choose which traits to favor. It then applies this same three-part framework directly to AI, arguing that most of its evolutionary influence will fall into the inadvertent category rather than deliberate genetic engineering.

To help readers think clearly about the many different ways AI interacts with human life, it proposes an ecological framework borrowed straight from biology. It describes AI relationships with humans using the same categories biologists use for species interactions, mutualism where both AI and humans benefit, commensalism where AI benefits without harming humans, parasitism where AI benefits at real cost to humans, predation where AI poses direct threats to human survival, and competition where AI and humans compete for the same limited resources like energy. Social media, in his account, began as a mutualist technology that helped people connect, but has since drifted toward parasitism because platforms are now optimized to capture attention and engagement rather than simply serve user needs.

The final building block connects these ecological relationships back to human social evolution specifically. Brooks describes how humans underwent 'self-domestication', a process in which cooperative, less aggressive individuals were favored over violent ones, partly because groups could conspire against and remove dangerous members. He argues AI could either support or undermine this process depending on how it is deployed, supporting it through better matchmaking and cooperative institutions, or undermining it through surveillance that protects despots and algorithms that reward tribalism and conflict.

Three Critical Questions to Ask Yourself

• Can you explain the difference between deliberate, inadvertent and natural selection, and identify which category most current AI applications fall into according to Brooks?

• Do you understand why Brooks predicts that AI-powered dating apps could either intensify or ease mating competition depending on which gender predominantly engages with AI companions?

• Can you describe at least one mechanism by which an AI system built for a completely unrelated purpose, such as criminal sentencing or social media engagement, could still influence human gene frequencies over time?

Roadmap

Choose one AI product you or your organization currently uses or builds, a dating app, a social platform, a hiring tool, or a virtual assistant, and classify it using Brooks's ecological framework as mutualist, commensal, parasitic, predatory, or competitive with human interests. Guidance: Justify your classification with a specific mechanism from the paper, such as engagement optimization or attention capture, rather than a general impression.

Working individually or in a group, identify one unintended selection pressure a specific AI system might be creating right now, drawing on the paper's examples of dating apps, criminal justice algorithms, or social media engagement systems. Guidance: Focus on who the system helps find partners, avoid incarceration, or gain social advantage, since these are the direct levers Brooks connects to reproductive success.

Draft a short set of design principles a company building an AI companion or matchmaking product could adopt to reduce the risk of intensifying mating market inequality, based on the specific predictions Brooks makes about male and female engagement imbalances. Guidance: Reference the paper's finding that both extreme male-biased and female-biased engagement patterns carry distinct social costs, so your principles should address both directions.

The Bottom Line

This lesson makes a provocative claim that most conversations about AI and humanity get the timescale wrong, focusing on dramatic near-term catastrophes while missing the slower, cumulative process of natural selection quietly operating through dating apps, chatbots and sentencing algorithms.