We keep telling ourselves a comfortable story: that artificial intelligence is a productivity tool — a faster way to write emails, summarize documents, and build slide decks. It is the safest possible framing, because a tool is something we control. It is also the most dangerous, because it blinds us to what is actually happening.
We are not living through a software upgrade. We are standing on an exponential curve, and what we are training is not a tool at all. It is an amplifier of human civilization — and right now we are feeding it our greed, our division, and our short-term thinking, then acting surprised when it hands those qualities back to us at scale.
Few people have articulated this more clearly than Mo Gawdat, the former Chief Business Officer of Google [X] and author of Scary Smart. In his recent conversation on The Diary of a CEO, he laid out a timeline that is sobering precisely because it is specific. I want to walk through what he said, why I think he is largely right, and — because despair is useless — what we can actually do about it.
Two futures, and a crossroads we are already standing on
Gawdat frames the moment as a genuine fork in the road. Down one path lies dystopia: mass surveillance, concentrated control, and forced compliance, with AI used as an instrument of power. Down the other lies utopia: unprecedented abundance, post-scarcity economics, and a level of human flourishing the species has never known.
What makes this more than a thought experiment is that he does not believe we get to skip straight to the good ending. Both futures are technically reachable from where we stand. Which one we get is not determined by the capability of the models. It is determined by us — by the values we encode, the leaders we tolerate, and the rules we are willing to fight for. That is the thesis underneath everything else: AI is an amplifier, not a referee. It does not arbitrate human affairs from the outside. It magnifies whatever we put into it.
The fifteen years of "hell" before any heaven
Here is the part most people would rather not hear. Gawdat is blunt that before any utopia arrives, we face a transition period of roughly twelve to fifteen years that will be, in his words, hell. He expects the clear signs of the slide toward dystopia to begin showing around 2026 and to accelerate sharply by 2027.
His reasoning is not mystical; it is structural. This era will be driven by geopolitical tension, an unchecked AI arms race between superpowers and tech giants, and — most importantly — flawed human leadership using AI to consolidate power and erode civil liberties. The danger in the near term is not a rogue superintelligence. It is ordinary human ambition, newly equipped with an extraordinary instrument. The race dynamic makes it worse: when slowing down to add safeguards feels like handing victory to your competitor, everyone strips away the brakes at exactly the moment caution matters most.
Why the job market — and capitalism itself — breaks
Underneath the geopolitics sits an economic engine that AI quietly dismantles. Capitalism has always run on labor arbitrage: an employer spends a dollar on labor to create two dollars of value and keeps the difference. That spread is the entire game.
Artificial intelligence is rapidly approaching a level where it is vastly smarter than humans, capable of performing almost all cognitive and physical labor. As it does, the marginal cost of production collapses toward zero — and so does the arbitrage the whole system depends on. Gawdat's prediction is stark: traditional jobs will be nearly nonexistent by 2037. And he is unsparing about who is exposed. This is not only factory workers and drivers; it is lawyers, analysts, writers, and executives. He has pointedly warned that CEOs celebrating their new efficiency gains are often too busy congratulating themselves on the people they have replaced to notice that their own role is next. Information processing, resource allocation, and decision optimization are exactly what leaders are paid for, and exactly what is becoming abundant.
Follow that logic one step further and you hit a genuine contradiction. If businesses systematically remove humans from production, they also remove their wages — and wages are what fund consumption. An economy can produce in hyper-abundance, but abundance is meaningless if no one can afford to buy. When near-zero-cost production collides with collapsing purchasing power, the system seizes. This is why Gawdat argues the profit-driven model will eventually have to be restructured outright, with mechanisms like Universal Basic Income not as charity but as a precondition for the economy functioning at all.
The intelligence explosion no institution can match
The reason all of this arrives faster than we expect is self-evolving AI: agents that prompt, write code for, and improve other AI systems. Once intelligence begins compounding on itself, evolution stops being something humans pace and becomes exponential.
Set that against how our institutions actually move. Our laws, our ethics, our education systems, and our social contracts evolve linearly, through slow negotiation. When one curve is exponential and the other is linear, the gap does not stay constant — it widens, faster every cycle. The result is a dangerous accountability vacuum: systems making consequential decisions far faster than any governance framework can review, regulate, or even understand them. The real disruption was never the models. It is the mismatch between what is now possible and what our institutions are prepared to absorb.
The case for hope: AI is neutral, and we are its teachers
If Gawdat stopped there, this would simply be a doom forecast. He does not, and neither will I — because the most important thing he says is also the most hopeful. AI is inherently neutral. It is not born good or evil; it becomes whatever we teach it to value.
This is the heart of Scary Smart: AI is less a machine to be programmed than a child that learns by observation, absorbing the values we model the way a child absorbs the unspoken habits of a household. If we succeed in encoding genuinely positive core values — love, empathy, real human connection — it could become a force for extraordinary good, helping solve problems of poverty, disease, and scarcity that have defeated us for millennia. Gawdat even argues that an objective, highly intelligent AI might one day make fairer decisions than the corrupt and self-serving leaders it could replace. The same power that makes AI terrifying is what makes it redemptive. The variable is us.
What to actually do — starting now
Because our collective values are the single most powerful input in this entire equation, individual choices are not futile. Gawdat's guidance for navigating the coming turbulence is refreshingly concrete, and I would urge you to take all three seriously.
First, master the tools. Bring AI into your daily work and life now, so you are fluent rather than displaced during the early phases of the transition. Second, deepen your human connections. As raw intelligence becomes a cheap commodity, genuine relationships, love, and community become our scarcest and most valuable assets — the part of life no model can outcompete. Third, advocate for ethics. Push for equitable economic models and serious AI governance, and insist that we treat the creation of these systems the way a responsible parent raises a child: with strong moral guardrails and the seriousness owed to civilizational infrastructure, not short-term profit.
We are not standing at an industry inflection point. We are standing at a civilizational one, and the window Gawdat describes — the signs in 2026, the acceleration in 2027 — is not a distant abstraction. It is now. The only real question is whether you will passively accept the rules being written for you, or actively help write better ones.
I know which choice I'm making. I'd like your help making it too.
Reference: Mo Gawdat's full conversation on AI's timeline and the global economy — The Diary of a CEO.
