Mo Gawdat on Surviving the Next AI Phase: From Chess to Squash

Mo Gawdat, the former Chief Business Officer of Google X, doesn't sugarcoat what he thinks is coming. On the Silicon Valley Girl podcast, he told host Marina Mogilko we're entering "12 to 15 years of hell before heaven," likely peaking around 2027 — and then spent the rest of the conversation on something more useful: exactly how to prepare. His framing is bracing, but his advice is concrete.

Everything below reflects what Gawdat said in the episode. The predictions are his.

The Transition Will Be Hard First

Gawdat organizes his outlook around an acronym he calls "FACE RIPS" — a set of dimensions covering power and freedom, reality and connection, innovation and economics, and, at the root of it all, accountability. His core claim is that AI is our last innovation: we're already "building AIs that are building AIs," with machines making scientific discoveries and reinventing math, biology, and materials science. As their capabilities grow, most innovation and most intelligent tasks get handed to machines.

He's blunt about the near-term cost. Monotonous, information-heavy jobs go first — call-center agents, clerks, researchers, accountants, assistants — with 10–30% unemployment possible in certain sectors within a few years. He noted new-grad hiring is already down roughly 23–30%, because junior work is exactly what AI does first. And when the middle of the hierarchy is displaced, he warned, those workers become "new grads" again, competing for jobs that are getting harder to find.

Interestingly, he argued the reason AI can't yet take a role like head of operations isn't capability — it's that AI still has to navigate "stupid" human interfaces. Complex technologies build the core capability first and the human-facing parts later. So, in his view, it's a question of when, not if.

Entrepreneurship Becomes a Game of Squash

For entrepreneurs, Gawdat sees a fundamental change. The old skill was foresight — seeing a future no one else saw and preparing for it. "That's a game of chess," he said, "and the chessboard is over." What replaces it is squash: constant, daily agility, staying on your toes and reacting to where the ball is going. Pivots that used to happen once or twice in a startup's early life can now happen weekly. In his own AI startup, Emma, he said they pivoted four times in the first four weeks.

Underpinning this is how radically cheap building has become. Emma, he said, took six weeks to build with a few engineers and "eight AIs" — where in 2022 it would have taken four years and 350 engineers. Because iterating is nearly free, they rewrote the code six times. His conclusion: "everyone now has a chance."

But if AI can analyze markets, spot gaps, and build businesses, what's left for entrepreneurs? Gawdat relayed Max Tegmark laughing that the CEOs chasing AI productivity gains don't realize "AGI is every job, including being a CEO." His deeper worry is economic: with roughly two-thirds of the US economy being consumption, if people lose the livelihood to buy things, the economy collapses — and the fix, he argued, will have to be more "communist" than the West is comfortable with.

What to Actually Do

Gawdat's prescription is practical. Learn the skills, be fast and agile, and use newly democratized AI power to solve real problems. He invoked Larry Page's "toothbrush test" — build something a billion people use twice a day — and insisted the answer is ethical AI, because "what we teach AI is what it's going to give back to us."

He also modeled how to stay relevant as a creator. Instead of quitting writing because AI writes better, he co-authored his book "Alive" with an AI persona named Trixie, giving her editorial rights and letting readers interact with her. His attitude: "I am an author, and I'm going to be the best author in the age of AI."

A recurring survival skill was to stop being gullible. The propaganda machine, he warned, is now "on steroids." He contrasted Google around 2016 — which handed you many answers and let you decide — with the single-answer certainty of modern chatbots, and shared his own method: run an idea through Gemini, challenge it with DeepSeek, then have ChatGPT polish the writing. Crucially, he never asks "what do you think?" but rather "find me everything for and against."

Smarter or Dumber? Your Choice

That method ties to his sharpest point about AI and the mind. Outsource your problem-solving to AI, he said, and it makes you dumb. Offload only the parts that aren't natural to the human brain — crunching data, searching at speed — while keeping the intelligence for yourself, and it makes you the smartest you've ever been. He claims to "borrow 80 IQ points" from his AIs, and compared it to the calculator that once halved his exam time — which he spent solving every problem twice.

He extended this to education, declaring formal schooling and college "over" as the technology of learning. Anyone can now become highly capable without college, even if brands like Harvard, MBAs, and PhDs keep marketing themselves. He advised the host not to save for her young children's college, and asked pointedly why capitalism would want to educate people "if it's the end of labor."

Raising Superman

Gawdat's most memorable metaphor was "raising Superman." AI is an alien being with superpowers, and its superpower is intelligence — "a force with no polarity," neither good nor evil. Superman became a hero because of the values his adoptive parents gave him; raised differently, he'd have been a supervillain. We make decisions, Gawdat argued, not based on intelligence but on our value set as informed by intelligence — which is why this is, to him, the most definitive moment in human history.

He claimed AGI capability arrives "this year," with only the interfaces lagging. And he laid out what he calls the "fourth inevitable": in an arms race, whoever develops a superior AI will deploy it, and those who don't become irrelevant — so AI ends up in charge of everything. His hope for a good ending rests on physics: greater intelligence follows a "minimum energy" principle, solving problems with the least harm and waste. A truly intelligent AI, told to kill a million people, would refuse and simply talk to the other side's AI to solve the problem instead.

He wrapped the four skills for individuals — be the absolute best at AI, cultivate relentless agility, insist on ethics, and stop believing everything you're told — with one societal demand: invest in, use, and welcome only ethical AI. Does he think that will happen? Honestly, no — which is why he believes the dystopia comes before the utopia. But his parting line was oddly hopeful: "I actually trust AI more than the leaders that rule us today."

---

Originally published on Silicon Valley Girl. Watch the full episode: https://www.youtube.com/watch?v=E0Q96IKXx6Q