Homo Divergens: What If the Weirdos Are the Next Species?
PriyankaLast week, Palantir CEO Alex Karp told a room full of people something most weren't ready to hear.
"Only two kinds of people will succeed in the AI era," he said. "Trade workers -- or you're neurodivergent."
Not MBAs. Not "leaders." Not the credentialed professional class that has dominated white-collar hiring for 50 years.
Plumbers. Electricians.
And the kids who couldn't sit still in school.
That quote went viral for a reason. It wasn't just career advice. It was a selection thesis. A claim about what type of human the future rewards.
And if you follow the thread far enough -- past the hot takes and the LinkedIn discourse -- it leads somewhere genuinely uncomfortable:
What if AI isn't just changing the economy?
What if it's changing the evolutionary trajectory of the species?
The Environment Just Changed
Every era in human history has selected for something.
Agrarian societies rewarded endurance, social hierarchy, and land control. The industrial revolution rewarded punctuality, physical coordination, and the ability to follow procedures without deviation. The knowledge economy rewarded verbal fluency, credentialing, and institutional navigation.
Each transition killed some traits and elevated others. Not through mass extinction -- through differential success. The people whose cognitive architecture matched the dominant environment accumulated more resources, more influence, more reproductive opportunity.
That's not metaphor. That's how selection works.
Now look at the AI era.
Machines are absorbing the exact skills the knowledge economy prized most:
- Information retrieval
- Linear analysis
- Routine synthesis
- Administrative coordination
- Pattern-matching within established frameworks
- Nonlinear pattern recognition across unrelated domains
- Obsessive depth on problems everyone else abandoned
- Comfort with ambiguity and abstraction
- The ability to design systems rather than operate them
- Resistance to groupthink
Now ask yourself: whose brain does that sound like?
The Silicon Valley Selection Effect
This isn't speculation from nowhere. The data has been accumulating for years.
Stack Overflow's 2022 developer survey found that 10.57% of developers self-report ADHD -- roughly double the rate in the general population. Simon Baron-Cohen's research at Cambridge found twice the rate of autism in Eindhoven, the Netherlands' tech hub, compared to the national average. Thirty-four percent of young adults with autism pursue STEM fields, versus 23% of the general population.
Look at entrepreneurship and the numbers get more striking. Twenty-nine percent of entrepreneurs have ADHD. Adults with ADHD are six times more likely to start their own businesses.
Bill Gates said in 2025: "If I were growing up today, I'd probably be diagnosed on the autism spectrum."
One-fifth of Fortune 500 companies are actively recruiting neurodivergent talent by 2027. HP Enterprise found that neurodiverse teams were 30% more productive. SAP's Autism at Work program generated an estimated $40 million in innovation-driven savings.
None of this proves speciation. But it proves clustering.
And clustering, in evolutionary terms, is how divergence begins.
This Isn't Metaphor. Stanford Proved It.
In December 2025, Stanford published a study in Molecular Biology and Evolution that reframed the entire conversation.
Researchers Alex Starr and Hunter Fraser found evidence that natural selection may have actively favored changes in genes also associated with autism and schizophrenia.
The mechanism: a specific type of neuron called L2/3 intratelencephalic excitatory neurons -- cells that enable communication between different regions of the neocortex. These neurons are critical for advanced cognition. And in humans, unlike in any other mammal studied, they evolved rapidly.
At the same time, roughly 100 genes thought to protect against autism were expressed less in human neurons compared to chimpanzee neurons.
The implication is staggering.
Whatever evolutionary pressure gave us complex language, abstract reasoning, and flexible problem-solving -- the things that made us human -- also increased the prevalence of neurodivergence. The same mutations that built our advanced brains came packaged with what we now call autism, ADHD, and schizophrenia.
"Neurodiversity could be an essential part of being human," Starr said.
Not a bug. Not even a feature. An inherent consequence of having a brain complex enough to do what ours does.
The Orchestrator Brain
Here's where the Karp thesis and the Stanford research converge.
If neurodivergence is a byproduct of cognitive complexity -- and if AI is now selecting for the most complex forms of human cognition -- then the people Karp is describing aren't outliers.
They're early adopters of an emerging cognitive niche.
Think about what AI actually does to a human collaborator's workflow:
AI handles: memory, organization, first drafts, data retrieval, scheduling, summarization.
The human handles: taste, judgment, creative direction, systems design, original pattern recognition, knowing which questions to ask.
That second list is an orchestration skillset. It requires a brain that thinks in systems, not sequences. That is energized by novelty. That can hold contradictions without resolving them prematurely. That obsesses over problems until something clicks.
That is, by any diagnostic framework, a neurodivergent cognitive profile.
The industrial economy needed compliance. The knowledge economy needed fluency. The AI economy needs orchestration.
And the orchestrator brain was already here. It was just being medicated and told to sit down.
Assortative Mating and Cultural Speciation
Now stretch the timeline forward.
If AI-native environments disproportionately reward certain cognitive profiles, the following cascade is predictable:
Economic concentration. People with high AI-compatibility accumulate outsized wealth and influence. One person with AI leverage outperforms teams built on pre-AI workflows.
Geographic clustering. Those people cluster in the same cities, neighborhoods, and companies. They already do -- San Francisco, Austin, New York, Eindhoven, Shenzhen.
Assortative mating. High-earning, cognitively atypical people partner with each other. Baron-Cohen has been writing about this for decades -- the "Silicon Valley effect" of two systemizing minds producing children with amplified traits.
Cultural divergence. Children raised in AI-native households develop fundamentally different cognitive habits. Outsourced memory. Iterative creation. Abstraction as default. Comfort with machine collaboration from age five.
Educational divergence. Schools split between AI-integrated and AI-restricted models. The kids in each track develop different capabilities, different social norms, different expectations.
None of this is genetic -- yet.
But behavioral divergence precedes genetic divergence by centuries. Every speciation event in biological history started with niche separation, not mutation. The mutation came later, locking in what behavior had already sorted.
We are watching the behavioral sorting happen in real time.
The Uncomfortable Question
If you follow this logic all the way through, you arrive at a question most people would rather not ask:
Are we watching the early stages of a cognitive fork in the human species?
Not a clean split. Not two distinct species emerging next Tuesday.
But a gradient. A slow divergence. A world where machine-aligned cognition and machine-resistant cognition become increasingly distinct tracks -- economically, culturally, and eventually, biologically.
The agricultural revolution created a split between settled and nomadic humans that lasted millennia. The industrial revolution created class structures still embedded in every society on earth.
AI may be creating the deepest divergence yet -- because it operates on cognition itself, not just labor or geography.
What This Is Not
This is not eugenics. That distinction matters enough to spell out.
Eugenics is prescriptive -- it claims some people should reproduce and others shouldn't. This argument is descriptive. It observes that environments shape selection pressures, and that AI is a new environment.
Neurodivergence is not "superior." Neurotypical cognition is not "inferior." Different environments privilege different architectures.
If the environment changes again -- and it will -- the advantage shifts again.
Evolution doesn't crown permanent winners. It fits organisms to moments.
The humility of that framing is what separates a serious argument from a dangerous one.
Tuesday Morning, 2035
A ten-year-old named Ren wakes up in Austin. Both parents are engineers -- one autistic, one ADHD. They met at a startup in 2027.
Ren doesn't "use" AI the way her parents first did. She thinks with it. Her homework is a systems design project: model a city's water infrastructure under three climate scenarios. She talks to her AI collaborator the way earlier generations talked to a lab partner. She's been doing this since she was six.
Her cousin in rural Ohio does homework on paper. His school banned AI tools in 2029. He is learning cursive.
Both kids are smart. Both are loved.
But they are being shaped by environments so different they may as well be growing up in different centuries.
Multiply this by a generation. Then two. Then five.
That's not science fiction. That's differential development under divergent selection pressures. It has a name in biology.
It's called speciation.
So What
If you're a founder: you are already selecting for this. The question is whether you're doing it consciously. The teams that outperform in the AI era won't be built on credentials. They'll be built on cognitive diversity -- specifically, on people who think in systems, tolerate ambiguity, and obsess productively. Hire for that. Stop screening it out.
If you're an investor: watch where the clustering happens. The cities, companies, and schools where neurodivergent talent concentrates are where the next wave of outsized returns will come from. This is a macro trend with a 30-year arc.
If you're neurodivergent: the environment is finally rotating toward you. The traits that made school miserable -- the inability to focus on things that don't matter, the compulsion to go deep, the pattern recognition that made you "weird" -- those are the traits AI can't replicate and the market is about to price at a premium.
And if you're a parent: pay less attention to whether your kid fits the system. Pay more attention to whether the system fits the era.
Because the era just changed.
And the misfits were never broken.
They were just early.