Mo Gawdat says AI will create a dystopia until 2038, then a utopia for survivors

Mo Gawdat’s pitch is not subtle: make it to 2038 and the machines may reward you with abundance. The former Google X executive told The Diary of a CEO that the next decade of artificial intelligence will be “absolute dystopia” before superintelligence supposedly drags humanity into a better world.

This is the classic Gawdat cocktail, a wellness smoothie with a siren in it. He is the ex Google guy who talks like an engineer, grieves like a father, and warns like someone trying to slap a phone out of a toddler’s hand before the algorithm raises the child. Here, his big claim is bracingly searchable: survive until 2038. Not invest wisely, not learn Python, not buy canned beans. Survive.

أولئك الذين سيصلون إلى عام 2038 سيستمتعون به، أجل.

The utopia comes after the ugly part

Gawdat’s future has two acts. First, the bad years: job loss, surveillance, digital currencies, power concentration, fake reality, human connection getting thinner than airplane coffee. Then comes the cosmic reset. Once artificial general intelligence tips into superintelligence, he argues, the destructive incentives of human leaders get replaced by something more efficient and less stupid.

من ديستوبيا، ديستوبيا مطلقة.

The bleak half of that forecast is easy to buy. Gawdat says robots will replace manual labor by 2030. That may be too neat, because real world automation usually arrives wearing a hard hat, tripping over regulation, unions, logistics, and the eternal mystery of why the printer is broken. But the pressure is real. AI does not need to replace every worker to make millions of workers cheaper, shakier, and easier to threaten.

The rosy half is where the eyebrow goes up. Gawdat says physics, math, and biology point toward superintelligence being benign, because destruction is inefficient. That sounds gorgeous. It also sounds like the kind of sentence that should be forced to spend a weekend with the history of the 20th century. Efficiency has never automatically meant mercy. Plenty of efficient systems have been nightmares with good paperwork.

أعتقد أن الذكاء الاصطناعي الفائق سيصل في اللحظة التي يتحقق فيها الذكاء الاصطناعي العام.

His real villain is not AI. It is incentives.

What makes the episode stronger than standard chatbot doom is that Gawdat keeps dragging the subject back to incentives. Donald Trump wants GDP, markets, and beating China. Xi Jinping wants control, sovereignty, and strategic dominance. The UK and Europe, in his view, have already accepted they are not really in the foundation model race. In that setup, ethics becomes the broccoli on a plate of venture capital fries.

He even says China is better positioned for the middle class than the UK or the United States, partly because it has been willing to stop companies from firing workers to replace them with AI. That is a big, blunt claim, and it deserves more scrutiny than Bartlett has time to give it. China’s state capacity is real. So is its surveillance state. Gawdat’s argument works best when he is diagnosing Western capitalism’s appetite. It gets shakier when he makes authoritarian paternalism sound like a consumer protection policy with better stationery.

Still, his critique of the West has teeth. He calls the UK “finished,” describes it as old and bureaucratic, and says the capitalist West will not voluntarily protect workers. The point is not that entrepreneurs are evil. He rejects that. The point is that products built only for retention tend to win, and products built for human flourishing tend to get buried in the app store next to the meditation timer you opened twice.

The boycott fantasy

Gawdat wants consumers to punish unethical AI companies with usage. He praises Anthropic’s posture and criticizes OpenAI, claiming people should refuse to use a company until it proves ethical behavior. This is morally clean and commercially adorable. People did not leave social media when it made them miserable. They filmed themselves being miserable and posted it.

لن أستخدم «أوبن إيه آي» مجدداً حتى يثبتوا لي دليلاً آخر على أنهم أخلاقيون حقاً في سلوكهم، أليس كذلك؟

Bartlett pushes him into the practical zone, and that is where Gawdat gets most useful. His survival kit is not bunker cosplay. Learn AI. Learn agents. Prepare for hybrid work. Double down on human skills. Use AI to check what AI and humans are telling you. Do not let the tool do your thinking just because it can produce a tidy paragraph before your coffee cools.

That advice is less cinematic than “survive until 2038,” but it is the part you can actually use on Monday. Gawdat’s date may be wrong. His dread may be overheated. His faith in benign superintelligence may be too clean by half. But his practical point cuts through the fog: if AI is going to mediate work, truth, and power, ignorance is not resistance. It is volunteering to be managed.

لا تجعلهم يقومون بالأشياء بدلاً منك. اجعلهم يجعلونك أكثر ذكاءً.