Sundar Pichai Doesn't Understand Why You're Anxious About AI

The CEO of Google says humans aren’t evolved for this much change. He’s right about the conclusion and wrong about the diagnosis. The real reason is more specific, more personal, and more uncomfortable for the company that built the engine.

Sundar Pichai went on the New York Times “Hard Fork” podcast last week and said something that sounds profound. “Humans aren’t evolved to process that much change.”

It landed well. It was retweeted, quoted, clipped into short-form video. It had the cadence of wisdom: humble, measured, vaguely scientific. The CEO of one of the most powerful companies on earth, graciously acknowledging that the technology his company is building might be too much for the species that built it.

It’s also a hand-wave.

Not because it’s wrong. It’s technically correct. The human brain did not evolve to process information environments that change at the speed of a transformer model retraining. But “we’re not evolved for this” explains nothing. It’s the equivalent of a mechanic looking at a car that won’t start and saying, “Cars are complicated.” True. Unhelpful. And conveniently vague enough to avoid the question of whether the mechanic is the one who broke it.

Here’s what Pichai didn’t say, and what the data actually tells us: AI anxiety is not one problem. It’s four. And they’re hitting different people for completely different psychological reasons. Treating it as a single, monolithic “we’re not ready” is not just intellectually lazy. It’s strategically dangerous, because you cannot solve a problem you refuse to diagnose.

The Number That Should Worry Google

The New York Times poll that prompted Pichai’s response found that only 16% of people believe AI is mostly good. Thirty-five percent think it’s mostly bad. The rest are somewhere in the uncertain middle.

That’s not a technology problem. That’s a trust problem. And trust is not a single metric. It’s a composite of psychological variables that run through the deepest architecture of the human mind.

When a person says “I don’t trust AI,” they are not making a rational assessment of large language model capabilities. They are making a psychological appraisal, filtered through their primary motivational engine, processed through their characteristic cognitive style, and coloured by their specific distortion logic. The feeling of distrust is real. The reason behind it is different for every type of person who feels it.

Pichai’s “not evolved for this” framing treats AI anxiety like a universal reaction to speed. It isn’t. It’s a set of specific psychological disruptions, each with a different mechanism, a different trigger, and a different solution. And if you’re the CEO of the company that built the engine causing the disruption, you should probably understand the mechanics of what you’ve broken.

Engine One: The Connection Problem

A significant portion of the population is driven, first and foremost, by human connection. Not as a preference. As a psychological necessity. Their primary battery charges through interaction, through being seen, through the warmth of belonging to something that recognises them as a person rather than a data point.

These people are not anxious about AI because it’s fast. They’re anxious because it’s replacing the human signal. When a student phones a university and gets an AI chatbot. When a customer service interaction becomes a conversation with a model that has been trained to sound empathetic without experiencing empathy. When the algorithmic feed replaces the community forum. When the voice on the other end of the line is synthetic and the person on the other end knows it.

This is not a speed-of-change problem. This is a connection-erosion problem. The anxiety is the correct appraisal of a system that is systematically removing the human element from interactions that were previously human. The threat isn’t that the technology is too advanced. The threat is that it’s advanced enough to simulate connection without providing it.

For these people, the 35% who think AI is mostly bad aren’t reacting to progress. They’re reacting to loneliness packaged as convenience.

Engine Two: The Logic Problem

Another segment of the population processes the world through evidence, verification, and logical coherence. These are the people who read the terms and conditions. Who check the sources. Who want to understand the mechanism before they trust the output.

These people are not anxious about AI because it’s fast. They’re anxious because it’s opaque. A large language model produces an answer, and the answer may be correct, but the path to the answer is a black box. You cannot audit it. You cannot trace the reasoning. You cannot ask it to show its work in a way that survives scrutiny.

For a mind that requires provenance, that needs to verify the logic chain before endorsing the conclusion, this is not a convenience problem. It’s an epistemic integrity problem. The AI says “trust me,” and the evidence-driven mind says “show me why,” and the AI cannot. Not because it’s hiding something, but because the architecture doesn’t work that way.

These people aren’t afraid of the future. They’re afraid of a future where you’re expected to trust an answer you cannot verify. That’s not anxiety. That’s a reasonable epistemic objection dressed up as technophobia.

Engine Three: The Autonomy Problem

Then there are the people who run on freedom. The ones who need to choose, to explore, to make their own path, to hack the system rather than be hacked by it. Their primary drive is agency. The feeling that they are the ones making the decisions.

These people are not anxious about AI because it’s fast. They’re anxious because it’s deciding for them. The recommendation algorithm chooses what they see. The navigation app chooses which route they take. The AI assistant chooses which answer they receive. The content feed chooses which version of reality they inhabit.

Every one of these micro-decisions used to be theirs. Now it’s the algorithm’s. And the algorithm doesn’t explain its reasoning. It doesn’t offer alternatives. It doesn’t ask permission. It just optimises for engagement, which is not the same metric as autonomy.

For a mind that is psychologically allergic to control, this is not a technology problem. It’s a sovereignty problem. The anxiety isn’t about what AI can do. It’s about what AI is doing to the feeling of being the one in charge of your own decisions.

These are the people who will be first to reject AI products that feel controlling, and first to adopt AI products that feel like tools. The difference isn’t the technology. The feeling.

Engine Four: The Foundation Problem

And then there are the people who run on stability. The ones who need the ground beneath them to hold. Who build systems, maintain standards, honour commitments, and expect the infrastructure to be there tomorrow the same way it was there today.

These people are not anxious about AI because it’s fast. They’re anxious because it’s shifting the foundations. The job they trained for might not exist in five years. The industry they spent a career in is being restructured by models that cost a fraction of a human salary. The standards they maintained are being replaced by “good enough” outputs that nobody is checking. The institutional trust they relied on, the university, the bank, the government, the employer, is being eroded by systems that optimise for efficiency over reliability.

For a mind that is driven by security, this is not a progress problem. It’s a structural integrity problem. The anxiety isn’t about innovation. It’s about the foundations crumbling while the people who built the building are still standing in it.

These are the people Pichai’s “not evolved for this” framing misses entirely. They’re not struggling with the speed of change. They’re struggling with the direction of it. They can see that the infrastructure is being rebuilt by people who don’t value infrastructure. And they’re right to be worried.

Why This Matters

If you treat AI anxiety as a single problem, you get a single solution. “Communicate better.” “Educate the public.” “Build trust through transparency.” These are the responses of people who haven’t diagnosed the illness. They’re prescribing paracetamol for four different diseases because they all present with a fever.

The connection-driven human needs reassurance that the human element isn’t disappearing. The evidence-driven human needs transparency into how AI systems reach conclusions. The freedom-driven human needs the ability to choose, override, and customise. The security-driven human needs proof that the foundations will hold.

One message will not serve all four. In fact, the message that reassures one type will actively alarm another. “AI will handle the routine tasks so you can focus on what matters” sounds like freedom to the agency-driven mind and sounds like unemployment to the stability-driven mind. “AI is getting more human” sounds like progress to the connection-driven mind and sounds like a threat to the evidence-driven mind.

Pichai’s framing, “we’re not evolved for this,” is the corporate version of prescribing paracetamol. It acknowledges the fever without diagnosing the cause. It sounds humble while avoiding specificity. It lets the company that built the engine express sympathy for the people being disrupted by it without actually understanding the mechanism of the disruption.

The Uncomfortable Question

Here’s the part Pichai didn’t address.

Google didn’t just build AI. Google built the feedback loop. The recommendation algorithm. The search engine that trained two decades of human information-seeking behaviour. The ad platform that monetised attention. The feed that learned which psychological buttons to push for each user and pushed them, relentlessly, for years.

The AI anxiety people feel today didn’t start with ChatGPT. It started with the algorithmic feed that spent a decade training each person’s brain to process information through a narrower and narrower psychological lens. The connection-driven human was fed an ever-more-social reality. The evidence-driven human was fed an ever-more-curated dataset. The freedom-driven human was fed an ever-more-rebellious narrative. The security-driven human was fed an ever-more-anxious worldview.

Google didn’t just build the AI that people are anxious about. Google built the algorithmic environment that amplified each person’s specific psychological architecture until the anxiety became personalised, intense, and, for many people, overwhelming.

Saying “humans aren’t evolved for this” is technically accurate. But it conveniently omits the fact that the evolution was accelerated by the very platform asking for patience.

The Real Diagnosis

The 16% who think AI is mostly good are not the optimists. The 35% who think it’s mostly bad are not the pessimists. They are four different groups of people, running four different psychological engines, each experiencing a different disruption, each needing a different response.

Pichai says the next generation will “rise to the challenge.” Maybe. But rising to the challenge requires understanding what the challenge actually is. And the challenge is not “too much change, too fast.” The challenge is that the most powerful information system ever built is actively amplifying the psychological architecture of every person who uses it, creating personalised anxiety profiles that feel like truth, and the company that built it is describing the problem in terms too vague to solve.

The algorithm that built itself is not just an AI problem. It’s a human problem. And it requires a human solution, not a technological one. The first step is not to communicate better or educate the public. The first step is to understand that the anxiety is not irrational, not uniform, and not a single thing.

It’s four. And until the people building the engine understand the mechanics of the people it’s running over, the 16% will not grow. It will shrink.

David Chadderton spent his twenties and thirties teaching people how to make life-or-death decisions at forty thousand feet. He now applies the same principles to consumer psychology, which, depending on the brief, can feel equally high-stakes. He’s the creator of the STAR Framework and the author of The STAR Framework: Rewriting the Rules of Consumer Engagement (NYC Big Book Award 2025), The STAR Operating System: Decode Mindset, Understand Motivation, Transform Human Behaviour, and Dear Algorithm, It’s Not Me, It’s You. By day, a Chief Marketing Officer. By night, a behavioural science obsessive who writes The Unoptimised Human because he can’t stop thinking about why people do what they do.