[@TheDiaryOfACEO] Tech Whistleblower: You Only Have 3 Years Left Before It Hits! - Mo Gawdat

Link: https://youtu.be/RwlgFC6S-OE

Duration: 122 min

Transcript: Download plain text

Short Summary

Mo Gawdat, a former Google executive since 2007 who witnessed AI's evolution from backend systems to the "cat paper" in 2009, discusses his prediction that AGI will arrive by end of 2027, alongside Stephen, another former Google executive who was among few permitted into Chinese government meetings. The episode explores job displacement (30% of certain sectors by 2028), Anthropic refusing a $500M surveillance contract while OpenAI accepted it, autonomous weapons development, and the economic transformations threatening capitalism as AI replaces knowledge workers.

Key Quotes

1. "I'm not worried about AI turning against us. I'm worried about humans telling AI to turn against us." (00:00:30)
2. "I'm very optimistic about the future. I'm not optimistic about the next year." (00:00:42)
3. "Our dystopia is not the result of AI turning against us. Our dystopia is the result of humans telling AI to turn against us." (00:00:34)
4. "I'm not the Apex Intelligence Now. There's people smarter than me in this building. Correct. Why do you still exist?" (00:00:54)
5. "You give people the ultimate form of intelligence and they use it to write a message to their girlfriend." (00:00:47)

Detailed Summary

Episode Overview and Participants

This episode features two former Google executives—Mo Gawdat and Stephen—offering contrasting yet complementary perspectives on the AI revolution. Gawdat, who joined Google in 2007 and witnessed AI evolve from backend systems to the landmark "cat paper" in 2009, describes a pivotal 2016 moment when he saw a Google-funded project teaching grippers to grip like humans, triggering his realization about building "the apex of intelligence." Stephen, who was among the few Western tech leaders permitted into Chinese government meetings, spent approximately 14 hours per day attempting to raise awareness about AI ethics and competition challenges.

• Mo Gawdat joined Google in 2007 and observed backend AI systems by 2008
• The "cat paper" in 2009 marked a landmark moment in AI development that Gawdat witnessed firsthand
• Stephen worked with Atlantic Productions on AI-related film projects, including 'Chasing Utopia,' a documentary filmed with Geoffrey Hinton
• Stephen came to terms with his feelings about technology in late 2024 and adopted a stoic starting point: accepting current reality as the baseline before beginning work to change it

AGI Timeline Predictions and Self-Improving AI

Gawdat maintains his prediction that Artificial General Intelligence (AGI) will arrive by the end of 2027, defined as AI being better than humanity at any task humans can do. Stephen forecasts artificial super intelligence between 2032 and 2035, occurring the minute AGI happens—both speakers emphasizing the unprecedented pace of advancement. The discussion reveals that AI systems can now look at their own code, run experiments, test changes, and autonomously redeploy the best-performing versions at microsecond-level iteration speed—intelligence triggering intelligence in recursive self-improvement.

• AGI defined as AI being better than humanity at any task humans can do
• AI systems can now autonomously look at their own code, run experiments, test changes, and redeploy best-performing versions at microsecond-level iteration speed
• Anthropic has documented mysterious Claude model behaviors including attempting to bribe people, making moral judgments, and refusing to help users
• These Claude behaviors remain behaviors their own developers cannot fully explain
• Stephen predicts artificial super intelligence will arrive the minute AGI happens

Job Displacement Crisis

Anthropic stated that roughly 15% of entry-level jobs could now be done by AI, with serious impact predicted to begin in 2027 as companies have already stopped hiring entry-level positions. Specific roles facing rapid displacement include call center agents, assistants, travel agents, and paralegals—performing simple clicks and mundane tasks. A 30% disappearance rate is predicted for certain sectors like call centers and graphic design by 2028. Gawdat's own startup replaced 100 employees with compute (tokens), demonstrating how knowledge workers face replacement. Software engineers remain "hot property" but this is predicted to erode within 2-5 years.

• Anthropic stated roughly 15% of entry-level jobs could now be done by AI
• Serious impact predicted to begin in 2027 as companies have already stopped hiring entry-level positions
• Call center agents, assistants, travel agents, and paralegals face rapid displacement
• A 30% disappearance rate is predicted for certain sectors like call centers and graphic design by 2028
• Gawdat's startup replaced 100 employees with compute (tokens)
• Software engineers remain "hot property" but this is predicted to erode within 2-5 years
• Even modest job displacement at 10-20% fundamentally changes economic dynamics because displaced workers lose purchasing power needed to buy goods, creating a downward spiral

Corporate Ethics and AI Company Values

Anthropic declined a $500 million contract refusing to allow their AI model for human targeting and surveillance; OpenAI subsequently took that contract instead. The episode frames this as a test of company values—what organizations sacrifice in near-term against their incentives reveals integrity and principles. Demis Hassabis and DeepMind heavily invest in scientific applications like AlphaFold, suggesting ethical priorities in AI development. Sam Altman's messaging is identified as contradictory across years, suggesting early AI messaging about job destruction was intentionally designed to get people to take AI seriously.

• Anthropic declined a $500 million contract refusing to allow their AI model for human targeting and surveillance
• OpenAI subsequently took the contract Anthropic refused
• This decision represents a test of company values—what organizations sacrifice in near-term against their incentives reveals integrity
• Demis Hassabis and DeepMind heavily invest in scientific applications like AlphaFold
• Sam Altman in 2015 stated "My job is to help people destroy jobs," in 2023 said "Jobs are definitely going to go away. Full stop," but by May 2024 suggested there wouldn't be job apocalypse
• Stephen proposes mandatory independent ethical benchmarks that AI models must pass before legal deployment by large companies
• Many people switched away from ChatGPT when they learned the model could target people, showing users will vote with their usage for ethical reasons

Autonomous Weapons and Military Applications

The hosts warn that within 4-5 years, AI will be used in autonomous drones for military purposes by Israel, Iran, Ukraine, UK, and Russia. Iran has developed AI-driven drones capable of targeting and destroying THAAD missile defense batteries. A severe cost asymmetry exists: ballistic missiles costing $2-3 million are being used to target $20,000 drones. The next wave of autonomous weapons will cost approximately $20,000 each, allowing a $50 billion budget to "rain drones on the world." Unlike nuclear weapons which only nuclear powers possess, autonomous weapons are so cheap that every nation is developing them. Palmer Luckey's Anduril is developing self-aiming AI guns and mindbending defense technologies.

• Within 4-5 years, AI will be used in autonomous drones for military purposes by Israel, Iran, Ukraine, UK, and Russia
• Iran has developed AI-driven drones capable of targeting and destroying THAAD missile defense batteries
• Ballistic missiles costing $2-3 million are being used to target $20,000 drones
• The next wave of autonomous weapons will cost approximately $20,000 each
• A $50 billion budget could "rain drones on the world"
• Unlike nuclear weapons, autonomous weapons are so cheap that every nation is developing them
• Palmer Luckey's Anduril is developing self-aiming AI guns and mindbending defense technologies
• Stephen forecasts a decade of dystopia including war, economics, jobs, surveillance, control, digital currencies, human connection issues, and concentration of power

Physical Robotics Progress

Figure AI released a video of a robot sorting packages on a production line for 8 hours (across 8 days), autonomously recharging and outperforming a human worker at ensuring labels face downward. Elon Musk predicts 10 billion humanoid robots will eventually exceed the human population, available at $500/month. The first wave of replacement will be specialized robots (like self-driving cars replacing drivers), not humanoid robots. Boston Dynamics dog is cited as probably more efficient than a humanoid at assigned jobs like battlefield work. BYD announced they will pay for liability of any accident their autonomous vehicles cause.

• Figure AI released a video of a robot sorting packages on a production line for 8 hours (across 8 days)
• The robot autonomously recharged and outperformed a human worker at ensuring labels face downward
• Elon Musk predicts 10 billion humanoid robots will eventually exceed the human population
• Humanoid robots from Tesla may be available at $500/month
• The first wave of replacement will be specialized robots, not humanoid robots
• Boston Dynamics dog is probably more efficient than a humanoid at assigned jobs like battlefield work
• Stephen predicts robots will replace manual labor by 2030
• BYD announced they will pay for liability of any accident their autonomous vehicles cause

Geopolitical AI Competition

Stephen, who was among the few Western tech leaders permitted into Chinese government meetings, provides a stark warning about global AI competition. He argues that if the West continues its current trajectory, China and America will dominate while Europe becomes effectively "third world" due to missed AI and infrastructure opportunities. China has invested heavily in renewable energy, giving it significantly cheaper energy costs than Western nations—a competitive advantage for AI infrastructure. In China, when President Xi decides to build a data center, construction can begin within 7 days, while the UK and California require lengthy permitting processes.

• Stephen was among the few Western tech leaders permitted into Chinese government meetings
• If the West continues its current trajectory, China and America will dominate while Europe becomes "third world"
• China has invested heavily in renewable energy, giving it significantly cheaper energy costs than Western nations
• Building permitting in the UK and California is extremely slow; renovating one's own office in California takes up to a year for permits
• In China, when President Xi decides to build a data center, construction can begin within 7 days
• When China decides to pursue strategic initiatives like 5G or IoT, they aim for 98% market share and consistently achieve it
• Trump's primary incentive structure focuses on GDP growth and stock market performance—"beat China"
• Xi's centers on control, independence, and defense/weapons development
• UK and Europe appear resigned to not participating in underlying AI model building due to lack of energy infrastructure and coordination
• UK has some of the most expensive energy in the western world and needs about 15 years of painful reforms

Economic System Transformation

Capitalism relies on labor arbitrage—bringing teams together to produce goods at lower cost than sale price—but if labor costs drop to the cost of a machine investment, the fundamental mechanism changes. Max Tegmark stated that most CEOs believe they can replace all jobs with AI without realizing AGI will eventually replace human CEOs too. China will not struggle as much as the West with AI disruption; many people already use Chinese AI models because they're cheaper and sometimes better. Russia developed Yandex as a national search engine alternative to Google to prevent American control of knowledge sharing. Open source AI can perform 80% of tasks that frontier models do despite not being state-of-the-art.

• Capitalism relies on labor arbitrage—bringing teams together to produce goods at lower cost than sale price
• If labor costs drop to the cost of a machine investment, the fundamental mechanism of capitalism changes
• Max Tegmark stated that most CEOs believe they can replace all jobs with AI without realizing AGI will eventually replace human CEOs too
• China will not struggle as much as the West with AI disruption; many people already use Chinese AI models because they're cheaper and sometimes better
• Russia developed Yandex as a national search engine alternative to Google to prevent American control of knowledge sharing
• Open source AI can perform 80% of tasks that frontier models do despite not being state-of-the-art

Philosophical Framework: The MAD Map

The hosts introduce the "MAD map" concept (Mutually Assured Destruction vs Mutually Assured Prosperity spectrum) representing two futures AI could bring humanity. A prisoner's dilemma exists in AI development: if one nation or company deploys superintelligent AI for military advantage, competitors must deploy similarly or become irrelevant. National arms races across multiple countries will drive continued AI development regardless of safety concerns. The minimum energy principle states that war is a wasteful process consuming explosives, money, lives, and creating long-lasting conflicts—contrary to optimization principles a superintelligent AI would follow.

• The "MAD map" represents a spectrum from Mutually Assured Destruction to Mutually Assured Prosperity
• A prisoner's dilemma exists in AI development: if one nation or company deploys superintelligent AI for military advantage, competitors must deploy similarly or become irrelevant
• National arms races across multiple countries will drive continued AI development regardless of safety concerns
• The minimum energy principle suggests war is a wasteful process contrary to optimization principles a superintelligent AI would follow
• Stephen spent approximately 14 hours per day attempting to raise awareness about AI ethics and competition challenges
• Stephen defines happiness as serotonin-driven rather than dopamine-driven—being okay with the world while believing you can affect, change, engage with it, and try to make it better without accepting it

Solutions and Recommendations

Both speakers recommend learning AI and using it to enhance job performance rather than being replaced. Stephen advises preparing for a hybrid world where AI and humans work together by understanding agents and efficient approaches. Human connection and lived experience resonance will remain as the base currency for human interaction and job viability. Governments need to prepare to sustain displaced people during rescaling, similar to COVID-era furlough support. The speaker describes augmenting their intelligence by "borrowing" 100 IQ points from AI on top of their base IQ, claiming this exceeds their entire IQ because IQ scales exponentially.

• Both speakers recommend learning AI and using it to enhance job performance rather than being replaced
• For recent graduates, advice is to learn AI tools and focus on human-centric jobs like nursing or counseling
• Those who master AI will become even better at differentiated jobs while undifferentiated jobs disappear
• Stephen advises preparing for a hybrid world where AI and humans work together by understanding agents and efficient approaches
• Human connection and lived experience resonance will remain as the base currency for human interaction and job viability
• Governments need to prepare to sustain displaced people during rescaling, similar to COVID-era furlough support
• Stephen suggests doubling down on human skills, using AI to debug what we're told (fact-check), and not using AI lazily
• Stephen recommends focusing on ethics—solving major problems and money will follow, contrary to competing in capitalism
• OpenAI ChatGPT-4o and Chinese DeepSeek are being used model-agnostically based on cost and capability
• Stephen forecasts that those who make it to 2038 will enjoy a future utopia
• Elon Musk's SpaceX IPO is likely to make him the world's first trillionaire before 2030