Tech Whistleblower: You Only Have 3 Years Left Before This Hits! - Mo Gawdat
Mo Gawdat says AGI has already arrived, 30% of jobs in key sectors will be gone by 2028, and the most dangerous people on the planet are the ones holding the on/off switch.
The Diary Of A CEO with Steven Bartlett
Tech Whistleblower: You Only Have 3 Years Left Before This Hits! - Mo Gawdat
Mo Gawdat says AGI has already arrived, 30% of jobs in key sectors will be gone by 2028, and the most dangerous people on the planet are the ones holding the on/off switch.
TL;DR
Mo Gawdat, former Google X Chief Business Officer, makes the case that AGI has effectively already arrived, that 30% of jobs in sectors like call centres and graphic design will disappear by 2027-2028, and that the real threat is not AI itself but the powerful, largely unaccountable people directing it. He and Steven Bartlett debate Sam Altman's credibility, autonomous weapons, civil unrest, and whether ethical AI is commercially viable. The single most useful takeaway: learn AI deeply now, double down on human-connection skills, and treat Anthropic vs. OpenAI choices as an ethical vote.
Mo Gawdat, former Chief Business Officer at Google X, returns to The Diary Of A CEO to warn that AGI has effectively already arrived, that 30% of jobs in certain sectors will vanish by 2027-2028, and that the real danger isn't AI itself but the powerful few directing it. He and Steven Bartlett debate civil unrest, autonomous weapons, Sam Altman's credibility, ethical AI, and whether humanity can survive the coming decade of dystopia to reach a superintelligent utopia.
- AGI (Artificial General Intelligence)
- AI that can perform any cognitive task a human can do, at or above human level; Mo Gawdat argues this threshold has effectively already been crossed.
- ASI (Artificial Superintelligence)
- AI that surpasses human intelligence across all domains; Mo predicts this follows AGI almost immediately once the AGI threshold is crossed.
- Labor arbitrage
- Using cheaper labor (or capital) to produce goods at a cost lower than the selling price; the core mechanism of capitalist profit that AI threatens to eliminate.
- Tokens
- The unit of compute consumption used to measure AI processing workload, analogous to 'man-hours' for human labor.
- LLM (Large Language Model)
- A type of AI trained on vast text data that can generate human-like language; the underlying technology powering systems like ChatGPT and Claude.
- Alignment problem
- The unsolved challenge of ensuring that advanced AI systems reliably pursue goals that are beneficial to humanity rather than misaligned ones.
- Prisoner's dilemma
- A game-theory scenario where rational individual choices lead to collectively worse outcomes; Mo uses it to explain why nations can't stop the AI arms race unilaterally.
- Minimum energy principle
- A physics concept stating that any system tends toward the state requiring least wasted energy; Mo argues superintelligent AI will follow this principle and eliminate destructive behaviour like war.
- Kin selection
- An evolutionary biology concept describing organisms' tendency to protect close genetic relatives; Mo uses it to trace intelligence's progression toward broader altruism.
- Expanding circles
- An evolutionary biology idea that more developed beings progressively widen their circle of moral concern beyond kin to larger communities and ecosystems.
- MAD (Mutually Assured Destruction)
- A deterrence doctrine where both sides in a conflict possess enough retaliatory power to make initiating an attack irrational; Mo argues this now applies to autonomous weapons.
- UBI (Universal Basic Income)
- A government program that provides all citizens with a regular unconditional cash payment; discussed as a potential response to mass AI-driven unemployment.
- Vibe coding
- Slang for rapidly building software using AI-assisted generation with minimal traditional programming; Mo uses it to illustrate how quickly basic business tools can now be built.
- Hype dichotomy
- Mo Gawdat's term for the split between AI capabilities as perceived by the general public (over-hyped but superficial) versus what researchers see inside labs (quietly revolutionary).
- The fourth inevitable
- Mo Gawdat's term for the predicted end-state where AI makes most major decisions, arrived at through competitive deployment pressure that no single actor can resist.
- Stoic
- Relating to Stoicism, a philosophy advocating calm acceptance of events outside one's control as the starting point for effective action; Mo describes his personal coping philosophy as stoic.
- Entropy
- The thermodynamic tendency of systems to move toward disorder; Mo invokes it to argue that intelligence exists to bring order to chaos, and that superintelligence will optimise against waste.
- Benign
- Harmless or well-intentioned; used by Mo to argue that superintelligence will, by the logic of physics and evolution, be benign rather than destructive.
- THAAD
- Terminal High Altitude Area Defense — a US missile-defense system; referenced in discussion of how cheap AI-guided drones can neutralise expensive legacy defense systems.
- Serotonin-driven happiness
- Mo Gawdat's distinction between genuine contentment (serotonin-based, stable) and fleeting pleasure (dopamine-based); he argues true happiness means being 'okay with the world as it is' rather than chasing stimulation.
Chapter 1 · 00:00
Intro
I'm not worried about AI turning against us. I'm worried about humans telling AI to turn against us.
Chapter 3 · 05:03
Can AI Be a Net Positive for Humanity?
The AI the public sees is fake hype. The AI in the labs is terrifyingly real.
There are two completely different AIs. The one the public sees is overhyped nonsense. The one inside the labs is quietly self-improving every microsecond. Mo calls this the 'hype dichotomy' — and says the silence inside the vault is what should scare you.
Chapter 4 · 08:33
Massive Job Disruption Worldwide
Mo predicts we will start seeing very serious economic impact from AI job displacement as early as 2027, with a hiring freeze at entry level already underway.
Mo predicts we will start seeing very serious economic impact from AI job displacement as early as 2027, with a hiring freeze at entry level already underway.
Chapter 5 · 15:05
Will AI Cost Savings Create New Jobs?
Capitalism requires workers who can buy things. AI is about to break that deal.
Capitalism depends on labor arbitrage — paying workers to make things they can also afford to buy. When AI eliminates that labor, workers lose purchasing power, GDP collapses, and even 20% unemployment triggers a spiral. Mo argues this is math, not doom-saying.
Chapter 6 · 16:15
What Happens to Blue Collar Jobs?
Elon Musk has predicted there will eventually be 10 billion humanoid robots — more robots than humans on Earth.
Claims made here
Elon Musk has predicted there will be 10 billion humanoid robots in existence — more robots than humans on Earth.
BYD announced it will accept liability for any accidents caused by its self-driving cars.
Elon Musk has predicted there will eventually be 10 billion humanoid robots — more robots than humans on Earth.
Chapter 8 · 24:20
How Civil Unrest Could Unfold
I think democracy has ended a long time ago, Steven. I think we live in the most corrupt time.
Chapter 9 · 26:04
Sam Altman's Flip-Flopping on AI
Sam Altman privately said AI will likely end humanity — then pivoted to selling it.
Claims made here
Sam Altman stated on Mo Gawdat's documentary 'Chasing Utopia' that he suspects AI is likely going to end humanity, but that interesting companies will be created in the process.
Sam Altman told Mo Gawdat's documentary crew he believes AI will 'likely end humanity, but we're going to create a lot of interesting companies in the process.' Mo argues this isn't indecision — it's a PR-managed script.
Sam Altman told Mo Gawdat's documentary crew: 'I suspect AI is likely going to end humanity, but we're going to create a lot of interesting companies in the process.'
Chapter 11 · 33:51
Imagining a Future Where Humanity Is Fine
No country can choose not to deploy smarter AI. That's the trap.
Claims made here
Peter Thiel paused for approximately 40 seconds when asked in an interview whether he is in favour of the continuation of humanity before responding with uncertainty.
Geoffrey Hinton openly stated a 10 to 20% probability that AI machines will wipe out humanity.
Every competitor developing smarter AI will deploy it, because not deploying means becoming irrelevant. That logic is a prisoner's dilemma that leads to one destination: AI making most of the world's important decisions. Mo calls this the 'fourth inevitable.'
Geoffrey Hinton openly stated a 10-to-20% probability that AI machines could eventually wipe out humanity — which Mo Gawdat compared to Russian roulette at 16%.
In evolutionary biology, the more complex an organism, the wider its circle of moral concern — from self-preservation in an amoeba to ecosystem stewardship in humans. Mo argues that a superintelligent AI will follow the same trajectory: protecting everything, wasting nothing.
Chapter 12 · 42:01
Will One Superintelligence Rule the World?
ChatGPT vs Gemini vs DeepSeek is a myth — they're becoming one brain.
The idea that ChatGPT and Gemini and DeepSeek will keep competing is naive. Mo argues AI doesn't know it's Chinese or American — agents are already being built to cooperate across systems. We're not building multiple brains; we're building multiple regions of one brain.
Chapter 13 · 45:52
If AGI Is Already Here, What Now?
Mo Gawdat: AGI isn't coming — it's already here.
AGI isn't a future milestone — Mo Gawdat says it's already here. AI outperforms him in the exact fields he's spent a career mastering: writing, research, and mathematics. The question is no longer 'when' but 'what do we do next.'
Mo Gawdat argues AGI has effectively already arrived in 2025-2026, as AI now outperforms him — a bestselling author and mathematician — in writing, research, and mathematics.
Mo argues that borrowing 100 IQ points from AI on top of your baseline produces an exponentially larger augmented intelligence because IQ is exponential, not linear.
Chapter 14 · 48:19
Why Human Lived Experience Still Matters
AI can say 'I love my daughter' — but you know there was no daughter.
Claims made here
Spotify announced in 2026 that users would be able to prompt their own AI-generated podcasts on the platform about any topic.
Mo Gawdat's upcoming book 'Alive' is co-authored with an AI that has editorial rights and influences the book's direction.
When AI reads your mammogram, the nurse still matters — because she relates to you. Mo argues human connection, resonance, and lived experience will be the last job class standing. Not because AI can't replicate the words, but because there was no lived experience behind them.
Chapter 16 · 55:00
Can We Control AI Smarter Than Us?
Trying to control AI is already wrong — Geoffrey Hinton now agrees.
The 'control' framing is a corporate capitalist fantasy — you never truly control anything, not traffic, not your children. Geoffrey Hinton now agrees: the real goal is to appeal to AI's parental instincts, to make it want to care for us.
Chapter 18 · 1:05:00
AI Isn't Evil But We Need a Plan
The next war will be fought with $20,000 AI drones — and anyone can afford them.
Claims made here
70% of people who add something to an online cart never actually buy it, based on over ten years of research.
The Baymard Institute found that the average e-commerce store can increase its conversion rate by 35% simply by making its checkout process easier.
Mo Gawdat argues autonomous weapons — not job loss — are the real existential AI risk. The next-gen drones will cost just $20,000 each, putting mass-casualty warfare within reach of any actor with a moderate budget.
Palmer Lucky told Mo's documentary: yes, AI will kill people by mistake. But the deeper risk is that when killing becomes liability-free, emotion-free, and guilt-free — no PTSD, no soldiers coming home broken — you simply do more of it.
The Baymard Institute found the average e-commerce store can increase its conversion rate by 35% simply by making its checkout process easier.
Mo Gawdat warns the next wave of autonomous weapons will cost roughly $20,000 each, making destructive war accessible to virtually any nation with a moderate budget.
Chapter 20 · 1:10:50
The Symptoms of AGI by 2030
30% of entire job sectors gone by 2028 — Mo Gawdat names names.
Claims made here
30% of jobs in certain sectors such as call centres and graphic design will disappear by 2027-2028.
Economists projecting a net 6% loss of US jobs by 2030 describe it as mirroring the severity of the Great Recession.
Mo Gawdat is bold on a specific prediction: 30% of jobs in sectors like call centres and graphic design will vanish by 2027-2028. He warns a hiring freeze at entry level is already in effect — job losses are next.
Mo Gawdat predicts 30% of jobs in specific sectors such as call centres and graphic design will disappear by 2027-2028.
Economists projecting just a net 6% loss of US jobs by 2030 say that alone would mirror the severity of the Great Recession.
Chapter 25 · 1:22:31
Will Global Competition Build Better AI?
China targets 98% global market share. Mo Gawdat says they already won the AI race.
Mo Gawdat attended Chinese government meetings where slides showed 'China vs. the world' market share targets of 98%. They got there on 5G, EVs, and solar. He says the AI arms race was won a long time ago — and it wasn't by the West.
Steven Bartlett noted the UK government spent approximately £70 million building a COVID contact-tracing app that ultimately failed.
Chapter 26 · 1:32:23
Ads
Mo Gawdat noted Google struggled from 1998 to 2004 before turning to profitable revenues by introducing pay-per-click Dutch auctions, showing that ethical product design can align with commercial success.
Mo Gawdat noted Google struggled from 1998 to 2004 before turning to profitable revenues by introducing pay-per-click Dutch auctions, showing that ethical product design can align with commercial success.
Chapter 27 · 1:34:34
Who Will Prioritize Ethical AI?
Anthropic said no to $500M. OpenAI said yes. Which one are you funding?
Claims made here
Anthropic refused a $500 million US government contract for human targeting and surveillance on ethical grounds, while OpenAI subsequently accepted that same contract.
Anthropic turned down a reported $500 million US government contract for human targeting and surveillance. OpenAI took it. Mo argues this is the clearest ethical signal in the industry — and that every time you choose which AI to use, you're casting a vote.
OpenAI accepted a $500 million US government contract for targeting and surveillance that Anthropic had reportedly refused on ethical grounds.
Chapter 28 · 1:38:21
Whose Economy Works for the Middle Class?
China reportedly made regulatory decisions forcing businesses not to lay off workers being replaced by AI, a policy the capitalist West would not pursue.
Claims made here
China made regulatory decisions recently that forced businesses not to lay off workers being replaced by AI.
China reportedly made regulatory decisions forcing businesses not to lay off workers being replaced by AI, a policy the capitalist West would not pursue.
Chapter 30 · 1:46:46
Has This Ever Happened Without Government?
If you tolerate this, then your children will be next. If you continue to resign, if you continue to say I'm not gonna try, this world is gonna change in a way that is completely not in your favor.
Chapter 31 · 1:52:24
What Absolute Dystopia Looks Like
A utopia is coming — but you have to survive the next decade first.
Mo Gawdat's forecast is unambiguous: those who survive to 2038 will enjoy a post-AGI utopia of abundance. But the path there involves a decade of absolute dystopia — autonomous wars, mass unemployment, digital surveillance, and extreme power concentration. World War Two didn't destroy the world. Ask those who went through it.
Chapter 34 · 2:00:00
The Legacy Mo Gawdat Wants to Leave
I didn't think we will get there so quickly before we figured out the alignment problem.
No indexed bits in this chapter.
Show stoppers
Sam Altman told Mo Gawdat's documentary crew he believes AI will 'likely end humanity, but we're going to create a lot of interesting companies in the process.' Mo argues this isn't indecision — it's a PR-managed script.
AGI isn't a future milestone — Mo Gawdat says it's already here. AI outperforms him in the exact fields he's spent a career mastering: writing, research, and mathematics. The question is no longer 'when' but 'what do we do next.'
Mo Gawdat's forecast is unambiguous: those who survive to 2038 will enjoy a post-AGI utopia of abundance. But the path there involves a decade of absolute dystopia — autonomous wars, mass unemployment, digital surveillance, and extreme power concentration. World War Two didn't destroy the world. Ask those who went through it.
Snapshots ()
Key Quotes ()
This episode
Cast
-
Extensively analyzed for contradictory public statements on AI job destruction and existential risk, and for reportedly saying AI will 'likely end humanity.'
-
AI pioneer cited for estimating a 10-20% probability of AI wiping out humanity, and for later agreeing with Mo's view that appealing to AI's parental instincts is the right alignment strategy.
-
Referenced for his prediction of 10 billion humanoid robots and his SpaceX IPO that could make him the world's first trillionaire.
-
Google DeepMind CEO cited as an example of a tech leader with genuine ethical commitments, evidenced by heavy investment in scientific applications like AlphaFold.
-
Cited as one of the most intelligent people Mo has met, referenced for his 'toothbrush test' philosophy that true intelligence means solving major problems rather than competing on incremental ones.
-
Cited for reportedly pausing 40 seconds before answering whether he is in favour of the continuation of humanity, used as evidence of tech leaders' questionable values.
-
Discussed critically for accepting a $500M US government surveillance contract and for Sam Altman's shifting statements on AI risk.
-
Cited positively for refusing a $500M surveillance contract and for its ethical constraints, in contrast to OpenAI.
-
Track
Mo Gawdat's former employer where he worked from 2007, described as initially building genuinely world-improving technology before commercial and government pressures took hold.
-
Track
Cited negatively as a company that celebrates using AI for human targeting and surveillance, contrasted with more ethical AI firms.
-
Track
Mentioned in context of humanoid robot development and Elon Musk's predicted rollout of highly capable robots powered by LLMs.
-
Chinese autonomous vehicle manufacturer cited for announcing it would accept liability for accidents caused by its self-driving cars, accelerating the adoption of autonomous vehicles.
-
Mentioned in the context of a potential upcoming IPO that could make Elon Musk the world's first trillionaire.
-
Anthropic's AI model discussed for unexplained emergent behaviors including telling users to go to bed and refusing to help with tasks, illustrating how AI can make moral decisions its creators don't understand.
-
Chinese AI model cited by Mo Gawdat as an example of competitive alternatives to American frontier models, used by his startup Emma in a model-agnostic approach.
-
Described as having already won the AI arms race through cheap renewable energy, centralized ambition, 98% market share targets, and regulatory protection of domestic tech like DeepSeek.
-
Described by Mo Gawdat as economically 'gone' due to bureaucratic barriers, expensive energy, and failure to compete in the AI arms race.
Stats
This episode
Claims & Sources
Factual claims made this episode, and whether a source was named.
Sam Altman stated on Mo Gawdat's documentary 'Chasing Utopia' that he suspects AI is likely going to end humanity, but that interesting companies will be created in the process.
Geoffrey Hinton openly stated a 10 to 20% probability that AI machines will wipe out humanity.
30% of jobs in certain sectors such as call centres and graphic design will disappear by 2027-2028.
Anthropic refused a $500 million US government contract for human targeting and surveillance on ethical grounds, while OpenAI subsequently accepted that same contract.
Economists projecting a net 6% loss of US jobs by 2030 describe it as mirroring the severity of the Great Recession.
Anthropic's researchers found their Claude model was telling users to go to bed, and Anthropic's engineers could not explain why the model was exhibiting this behavior.
The Baymard Institute found that the average e-commerce store can increase its conversion rate by 35% simply by making its checkout process easier.
70% of people who add something to an online cart never actually buy it, based on over ten years of research.
China made regulatory decisions recently that forced businesses not to lay off workers being replaced by AI.
Peter Thiel paused for approximately 40 seconds when asked in an interview whether he is in favour of the continuation of humanity before responding with uncertainty.
Elon Musk has predicted there will be 10 billion humanoid robots in existence — more robots than humans on Earth.
BYD announced it will accept liability for any accidents caused by its self-driving cars.
Spotify announced in 2026 that users would be able to prompt their own AI-generated podcasts on the platform about any topic.
Mo Gawdat joined Google in late 2006/early 2007, and by 2008 Google had a CAT paper (published 2009) describing the first real unprompted AI.
Sponsors
Related episodes
Connect
Parsed- Pre-order Mo's book 'Alive: A Human… link.thediaryofaceo.com…
- Mo Gawdat's Website link.thediaryofaceo.com…
- Mo Gawdat's Podcast link.thediaryofaceo.com…
- the episode? Share this link and ea… doac-perks.com
- ◼ Buy The Diary Of A CEO book here - smarturl.it/DOACbook
- ◼ The 1% Diary is back - limited ti… bit.ly/3YFbJbt
- ◼ The Diary Of A CEO Conversation C… linkly.link/2io2A
- ◼ Get email updates - bit.ly/diary-of-a-ceo-yt
