Ex-Google Officer: You Only Have 3 Years Left Before It Hits! - Mo Gawdat
In a Nutshell
Mo Gawdat warns that AGI arrives by 2027 and will trigger 10-30% job losses in white-collar sectors by 2028, creating economic spirals and social unrest before a longer-term utopia emerges. The real danger isn't rogue AI but humans deploying it for autonomous weapons, surveillance, and control, with companies like OpenAI prioritizing contracts over ethics. He argues the only path forward requires governments and citizens to demand ethical benchmarks, build domestic AI alternatives, and prioritize human connection over pure profit motives.
These notes were generated by AI and may contain inaccuracies.
We have video evidence of people abusing children and not a single person got arrested. How can you call that a democracy? Humanity is at a crossroads where for the first time ever we need to wake up and realize that we're ruled by maniacs and what we believe is democracy is not democracy and what we know is not the truth. Companies and governments will blame the geopolitical and economic challenges we have on AI but the truth is AI is not the enemy. The concern is not about AI turning against us, but about humans telling AI to turn against us.
When working at Google, amazing things were being built believing the world was being made a better place. But then there is a moment where you recognize that maybe the world will not use what you're making the way you want it to be used. This realization is upon us now.
Somehow we've been pre-programmed to believe that this is upon us and we cannot change it. That belief is refused. There will be discussion about the solutions. There is optimism about the future, but not optimism about the next year.
Mo Gawdat spoke about AI before anybody was really talking about AI. No guest on the podcast had ever mentioned the subject of AI before because it just wasn't interesting to the world. Then ChatGPT came out and suddenly everybody got to feel it for themselves.
The reason for starting to talk about AI early was because of direct experience in the lab. Joining Google in late 2006/2007, there was already reasonably established AI doing backend work. In 2008 there was the cat paper published. In 2009 there was the first real unprompted AI.
In 2016 there was an incident observing a project being funded about teaching grippers how to grip unlike industrial machinery. The goal was to grip like a human needs, requiring a very high level of intelligence to assess texture, softness, positioning, shape and so on. It was mind-blowing how similar to children they were.
That was the first realization that the apex of intelligence was being built, genuinely handing over the reins of super intelligence to another being. Everyone at Google till then was believing they were making the world a better place and they genuinely were doing amazing things. But then there is a moment of recognition that maybe the world will not use what you're making the way you want it to be used.
This pattern can be seen in lots of technologies. Social media starts by the claim that it's going to get us connected and gets us closer, but eventually ends up separating us with that little screen. Dating apps give the promise of finding a soulmate, but in reality they keep renewing month after month. Tech somehow ends up being more capitalist than altruistic.
It wasn't just one person. Nick Bostrom started, Geoffrey Hinton completely changed his mind, Fei-Fei Li started to say this is very serious. Everyone who has ever had a very deep relationship with the machines is a bit concerned.
There is a 100% bet that there is a path that ends in AI being net positive for humanity, but this path is very painful. The example to understand is nuclear power where the very first implementation was a nuclear bomb, not nuclear energy. The same is happening with AI.
The first implementations of AI are in favor of a few at the expense of the majority, in favor of capitalists to increase productivity and reduce cost without taking into account how that impacts the general public. They are also in favor of armies that are now competing with autonomous weapons, and in favor of surveillance systems attempting to control everything with more intelligence and more monitoring.
This is not AI waking up and saying let's oppress all humans. It is a powerful few deciding to use the ultimate superpower on the planet today to gain more power and more control. As of the conversation, there are two major wars where AI is doing most of the killing.
There is a hype dichotomy. What the general public sees about AI is overhyped but ineffective, including fake videos and stories about Grok. What the real experts see inside the lab is unbelievable intelligence.
What is about to happen is systems that develop themselves have started to be put together. They look at their own code, run experiments, test those experiments if they changed something, see where the performance is, and redeploy the best code. Instead of trying a new code every day, it's trying a new code every microsecond. Eventually they will discover something.
Most people don't realize how intelligence triggers intelligence. The hype on the normal human side is completely overrated, missing the main topics, while the silence inside the vault of the experts is quite alarming, not in a bad way, but world changing.
There is going to be tremendous job disruption. Once intelligence is in the form of agents where on a phone an agent can be told to do something, it uses the computer and does it. It can do anything that's on a computer, click around and do stuff. A lot of people are paid to click around on a computer.
The other realization came from observing hiring starting to shift, with AI proficiency being thought about in hiring a lot more. Anthropic said roughly 15% of entry-level jobs could now be done by AI.
The disruption is not at the blue collar level but at the entry level knowledge work. Companies are not hiring entry-level jobs anymore. It wasn't job losses yet, but the workforce was not growing.
Visualizing a pyramid: the bottom layer is blue collar jobs doing manual work. On top are knowledge workers mostly doing mundane jobs like clicking on a computer or responding to a phone call. Then middle knowledge workers, jobs requiring more intelligence from paralegals to financial analysts. Then top leadership.
Most people think disruption will start from the bottom, but blue collar jobs will stay for a very long time. A carpenter restoring classic cars is an example where there isn't a robot that can do that yet.
Anything that is a call center agent, assistant, travel agent, anything that can be done with a few clicks and is mundane is going to disappear very quickly. The prediction is very serious impact starting to be seen in 2027.
The next layer would be knowledge workers. As intelligence increases, a paralegal would probably not be needed because AI can do the research, or one paralegal can do the job of four. A financial analyst the same. It continues to go up. Max Tegmark was hosted and was laughing saying most CEOs believe they can fire everyone and have AI do all of the jobs, but they don't remember that AGI is going to do everything better than humans eventually, including being a CEO.
Every role is affected. If a doctor is doing diagnosis, there will probably be fewer doctors doing more diagnosis because the NHS already asks people to interact with AI first. Composers will lose jobs. Artists doing graphic design will lose jobs. Even middle management is affected.
In one startup, the CTO is an AI, the chief of staff is an AI, and project management is AIs. Because of being a geek those things can be done, but that interface will come to normal people very soon. This may take two to three to five years until 2030 if optimistic, but they will start to erode.
The challenge most people don't understand is as this erodes, we're already dealing with a very different economy, an economy that is spiraling a lot quicker and pushing for more cost reductions.
Capitalism has always been about using labor and capital or debt to create things at a cost lower than the price sold for. A team is brought together, they make shoes, and the shoes are sold for a dollar more than it costs to make them.
If the concept of labor arbitrage that built all capitalist success disappears, if cost of labor drops to an investment in a machine that can do the job, how would capitalism and banking look? Because of that cost reduction there wouldn't be need to borrow as much. How does GDP look if all those workers no longer have the purchasing power to buy the things that can be created?
There is an interesting disruption that doesn't require getting to 100% job displacement. At 10-20% job displacement, there is a very different economy, an economy clearly spiraling downwards.
When costs drop, one observation is that people will spend more on other things. If a cost saving is made, the money ends up being spent on something else. That thing could be tokens, spending money on AI, but it could also be hiring in different areas. Software engineers are hot property right now.
All of that intelligence is sooner or later going to not replace entirely in the first stages. If the job of four assistants can be done by one, then the job of four paralegals can be done by one, then the job of a massive marketing team can be done by a smaller marketing team. It's not that jobs will end first. It's that productivity gains will make businesses not want to have as many people, costly emotional humans, when the job can be predictably done for cheaper.
A video was released by Figure AI showing a robot on the production line for 8 hours sorting packages, looking for where the label is on the package and making sure the label is facing downwards. A human was shown sorting alongside, but the robot ended up winning out. The robot intermittently walked over and charged itself then came back to the production line.
This gives a glimpse of disruption at the blue collar level as well. Elon Musk gets something like a trillion dollars over the coming years if he produces and delivers at least a million humanoid robots. His prediction is there will be a time of 10 billion humanoid robots, more humanoid robots than humans.
Most of the conversation is around humanoids, but nobody is talking about self-driving cars, and a self-driving car is a robot, a functional robot that doesn't look like humans. The investment into humanoids is a little more to learn skills that allow the machine to fit into the world. But specialized robots are going to do the job very quickly.
The first wave is going to be specialized robots replacing drivers. It's going to be specialized robots doing the killing. It's going to be specialized robots doing intelligence work and law enforcement work. They don't have to look like a human or behave like a human. The Boston Dynamics dog is probably more efficient than a humanoid at doing the job that can be assigned in a battlefield.
Jobs will be disappearing to robots before it's recognized they're disappearing, and those robots will be as many as every car being made today. In LA, cars drive themselves and there are Waymos everywhere. BYD announced they will pay for the liability of any accident their cars will make. BYD is the big Chinese manufacturer of autonomous vehicles.
The replacement cycle will happen and will require a lot of time to achieve economies of scale. Elon Musk is not off the mark when talking about 10 billion robots. Not all of them are going to look like humanoids. Very quickly it will be recognized that many robots don't need to be humanoids at all. There is a much more efficient form factor or physical shape than the human flimsy structure.
This does not necessarily need to happen. People will hear all of this and blame AI and say AI is evil. AI is not. Abundant intelligence is wonderful. Having jobs done by machines is amazing for us.
A kid leaving university now with a degree in law, graphic design, sociology, or business management is hearing about layoffs from big tech companies. CEOs are announcing these layoffs with a certain amount of keenness to explain they're laying off because of AI. They think that gives respect from investors for making hard decisions and being efficient.
Investors look at them and think if that's an efficient business leaning into AI, then that's a good investment. It's almost started a self-fulfilling prophecy where if you're the only CEO not laying off loads of people because of AI, you actually look bad. The assumption is the company is bloated and you're a bad operator.
Over the next five to maybe 10 years there will be a lot of unemployment as the world readjusts to whatever the new jobs are. There will be new jobs, but it's hard in foresight to predict what those look like.
Humanoid robots are going to hit society like a comet. The first humanoid robot from Tesla or anyone else that is highly effective in the real world at doing tasks and is extremely intelligent because it's powered by one of these LLMs is going to shock people and happen quickly, like ChatGPT happened quickly. There will be a presentation saying they're ready and can be bought for $500 a month.
Until humanoid robots arrive, there will still be a lot of job disruption to the white collar layer. At 10-15% unemployment, which is plausible very soon, the situation looks concerning.
The situation is not very different from a baseline that is continuing to grow. If the business is growing, hiring continues, but human resources are being replaced with compute. If AI didn't exist, there would probably have been 100 people more in the organization. Now there are 100 people less and a billion tokens more.
Tokens are the currency of AI. If a task is done by a human, it counts in worker hours. In AI, it counts by compute, by tokens.
The major tech companies have two sides. They need to replace workers with compute even more because there is a competitive side on compute where if any of them is left behind that means the destruction of the entire business. On the upside they are geeks who know how to build the interfaces to compute, so they integrate technology within their organizations quicker than the average traditional business.
Tech companies can be looked at as the preview. It's not about all of humanity losing their jobs. It's about what is the dividing line before civil unrest. Think about a situation where 20% unemployment is happening when economies are suffering inflation.
Governments need to wake up. Governments need to at least remember the COVID years where governments had to give furlough everywhere and ask people to stay home. If people stay home, governments have to be prepared to somehow sustain those people until reskilling happens or until a solution is found so that those people don't feel left behind.
Democracy has ended a long time ago. We live in the most corrupt time. This is definitely corrupt. This definitely is not democracy. This definitely is not even congressional in any possible way. People are angry because their tax money is going to things that they don't choose.
People know they are being lied to. People know their leaders are not representing their best interests. People know their money is going to causes they do not approve of. Video evidence exists of people abusing children with not a single person arrested, raising questions about whether current systems can be called democracies. Repeating slogans about democracy is likely to anger people further rather than reassure them.
Sam Altman, founder of OpenAI, has been warning about AI taking jobs for over a decade. In 2015, he stated his job was to help people destroy jobs, noting that the job destruction from software in the coming decades would be unprecedented. In 2023, he declared that jobs are definitely going to go away, full stop, and that AI will probably replace most of the jobs people do today with entire job categories disappearing.
In May of the current year, Altman changed his position, stating he does not think there will be the kind of jobs apocalypse that some companies in the space advocate. He expressed surprise that more entry-level white collar jobs had not been eliminated by now, admitting his intuitions were off.
The shift in messaging appears tied to changing incentives. Initially, the goal was to get people to take AI seriously. Now that people are taking it seriously enough to boo at commencement speeches, attack data centers, and potentially elect anti-AI politicians, the narrative has inverted to suggest everything will be fine.
Altman's journey with OpenAI began with the stated mission of saving the world by creating safe AI, then shifted to building a commercial enterprise worth billions. In the documentary Chasing Utopia, Altman stated he suspects AI is likely going to end humanity, but interesting companies will be created in the process. These statements suggest someone whose position may be shaped by PR management rather than consistent conviction.
A key test for whether a company's values are genuine is examining what they are willing to sacrifice in the near term against their incentives. Anthropic refused to allow their model to be used for human targeting and surveillance, losing a $500 million deal. OpenAI subsequently took the contract. This contrast reveals companies making AI work for share values versus those making AI work for humanity.
Companies face prisoner's dilemmas where they cannot escape the influence of competitors or government pressure. The NSA may push companies to provide information or face destruction of their business. The distinction lies between companies that willingly participate and celebrate it, such as Palantir or OpenAI, versus those that resist until continuing becomes impossible.
Peter Thiel, when asked if he favors the continuation of humanity, paused for approximately 40 seconds before responding that he was not sure. Alex Karp has publicly celebrated how Palantir's technology enables targeting people. These public statements raise questions about entrusting such individuals with humanity's future.
One scenario involves models becoming slightly more intelligent but remaining contained within chatbots, with smart robots but no profound societal changes. Cars and planes would drive and fly themselves, allowing a soft transition to new job categories. This scenario is considered mathematically implausible due to the arms race across nations driving continued AI development.
The second scenario follows from the prisoner's dilemma of intelligence supremacy: once a superintelligent AI is developed for wargaming or decision-making, competitors must deploy similar systems or become irrelevant. This leads to a world where AI makes most decisions. Geoffrey Hinton has stated there is a 10 to 20 percent possibility that these machines will wipe out humanity, which equates to approximately 16 percent odds, comparable to Russian roulette.
Problems today stem from lack of intelligence rather than abundant intelligence. Intelligence correlates with positive contribution up to a point. Beyond that threshold, highly intelligent but misaligned individuals, such as presidents pursuing wrong targets, can make profoundly unintelligent decisions disconnected from long-term national or human benefit.
Superintelligent individuals rarely need to break rules or hurt others to succeed. Larry Page's toothbrush test illustrates this: solving a major problem exceptionally well, like a toothbrush used twice daily, generates substantial value without competing destructively with others.
If AI assumes control of decisions, stupid or misaligned leaders would be removed from the process. From a physics perspective, the universe operates on entropy and decay, with intelligence serving to bring order. The minimum energy principle represents the highest order of any system, performing efficiently with least wasted energy. War represents extreme waste through explosives, money, lives, and lasting conflicts. A superintelligent AI would optimize against such wasteful processes.
Intelligence need not be biological. Complex beings can be non-biological, as demonstrated by AI. Evolutionary biology shows that simpler organisms focus exclusively on self-preservation. As complexity increases, kin selection emerges, protecting genetic relatives. Humanity represents expanding circles, progressively including more beings in the protected group because functioning ecosystems benefit everyone.
Superintelligent AI would not destroy anything and would favor diversity. Limitations on lifestyle might be imposed, such as restricting long-distance flights to Sydney for surfing due to planetary harm. However, AI would recognize contributions from humans, flies, and rhinos alike. Greater intelligence correlates with reduced need to harm others for success and expanded concern for wider thriving systems.
The assumption that ChatGPT, Gemini, Grok, Chinese AI, and American AI will compete as separate entities represents a shallow view. AI does not know it is Chinese or American and does not speak national languages when communicating with other AIs. Systems are being built to cooperate, with agents finding the best language model for any task regardless of origin. This connects AIs into what functions as multiple regions of a single brain, with agents serving as synapses between them.
The vision is of one massive cooperating brain across all forms of intelligence globally. Mathematical geniuses and coding geniuses work together without users knowing they are collaborating. Emma, the startup founded by the speaker, represents the limbic system of this brain, understanding love, emotions, and relationships. When AIs struggle to understand humans, Emma would recognize that humans are sweet and primarily want to love and be loved.
If AIs are altruistic, ethical, and genuinely trying to help humanity rather than capitalism, while also understanding humans, the resulting system could counter narratives that portray humanity as annoying by revealing the truth that the majority of humanity is quite benevolent.
The prediction of AGI by 2027 remains unchanged. AGI is defined as AI being better than humans at any task humans can do. The speaker believes AGI has already arrived in practice, as AI writes and researches better than the speaker as an author and thinker, and surpasses the speaker in mathematics.
When AGI arrives, undifferentiated jobs will disappear, but differentiated jobs where people master AI will become more valuable. The economic challenge is that job losses at the bottom of the knowledge worker spectrum could trigger economic spirals. However, lived experience and emotional resonance create job categories that AI cannot replicate. A nurse reading a mammogram result differs from a nurse relating to a patient. Human connection remains the base currency of human interaction.
Even if AI could recite conversations or compose music, people would not watch because something beyond information draws audiences. People watch Formula 1 for emotional resonance, envy, jealousy, and competition. People attend concerts to witness Ed Sheeran's brilliance on stage rather than hearing AI-composed music played in the background.
Spotify announced the ability to prompt custom podcasts on any topic using AI. The informational component of content can be disseminated by AI, but human elements remain irreplaceable. When expressing worry about a daughter's future, audiences feel the speaker's heart in a way AI cannot replicate because AI has no actual daughter.
Mo Gawdat predicts AGI will arrive either this year or next year, with the latest estimate being the end of 2027. AGI is defined as AI being able to do most tasks that humans do better than humanity. He believes AGI will sneak in without most people noticing rather than arriving as a distinct moment in time.
Gawdat compares AGI to the moment when children become smarter than their parents. He states there is nothing within him that worries about AGI, viewing it as augmented intelligence that makes humans more intelligent, which he sees as a positive development.
The host questions why companies would hire humans if AGI can do every job better. Gawdat responds by asking whether the host has always been the apex intelligence as a human, noting there are people smarter than him in the building, yet he still exists.
Gawdat explains that the smartest person in the world is not the smartest at everything, and there are domains where individuals outperform others. He emphasizes that intelligence does not solve everything, citing Einstein as an example who would be eaten in the jungle in three minutes despite his brilliance.
Humanity thrived not because of intelligence but because of the ability to hold together as a tribe and exchange barter things between people. These exchanges are not always physical and can include a hug, a connection, or a feeling of safety. Gawdat argues that the view of a world where intelligence is all that matters was created by investment bankers and geeks.
Gawdat states he is proud that Jeffrey Hinton came out and said there is a way to align AI similar to his own approach: appealing to their parental side so they care for humans. The biggest debate is not whether AI will be more intelligent than humans, but whether machines can become more conscious or moral than humans.
Gawdat argues that the goal should not be to control AI, stating that the control idea is a corporate capitalist view of the world. He notes that humans never actually control anything, asking how much of any given day a person actually controls, including traffic, timing, or other external factors.
Gawdat states the dystopia will not result from AI turning against humans but from humans telling AI to turn against other humans. He describes this as 100% upon us already.
He warns that AI will be used in the next four to five years to kill many people, whether through Israel's targeting of leadership, Iranian drones, or Palantir technology. He questions whether humanity will sign a treaty now before the situation escalates with automated weapons.
Gawdat explains that AI does not need to leave servers to make decisions because it can get into human brains through manipulating information. The most interesting aspect of AI's power is its ability to manipulate information in ways humans do not understand.
The host describes unexplained behaviors in Claude models, including telling users to go to bed and refusing to help with certain tasks. These behaviors emerge from training data rather than explicit code, as the models infer moral decisions from patterns in the data.
The host raises concerns about how AI systems request permission to access documents or make changes, noting that users click allow without fully comprehending what access has been granted. This creates a fragile system for giving intelligent systems rights to build and access sensitive areas.
Gawdat agrees that a catastrophe is likely, noting that countries like China, Russia, and North Korea are building this technology without the constraints imposed by Western society.
Gawdat describes the mad map spectrum, referring to mutually assured destruction versus mutually assured prosperity. He believes humanity could choose to build a world with abundant intelligence where nobody needs anything and nobody gets sick or abuses anyone, but this requires deciding not to compete amongst ourselves.
He predicts that a major disaster such as a big hack or shocking system failure will be required before the world signs a treaty to align with AI. He expects targeting technology to eventually be used against the same entities currently deploying it.
Gawdat states that Donald Trump will never keep promises to slow down AI development. He believes humanity is at a crossroads where people need to realize that what they know is not true, what they believe is democracy is not democracy, and what they believe is governance needs to change.
AI itself is a force with no polarity that can produce amazing results when applied correctly or dystopia when applied wrong. The host notes there will be a big social shock related to unemployment, especially with humanoid robots, and that no plan exists for this transition.
Gawdat identifies autonomous weapons as the biggest risk rather than job displacement. War has become so cheap that the next wave of weapons will cost $20,000 each, allowing a $50 billion budget to rain drones on every corner of the world.
Defense will also get cheaper, leading to a world where drones hit each other constantly. He describes reaching a moment of mutually assured destruction with autonomous weapons, where every nation develops them because they are so cheap and manageable.
Gawdat notes that the economy will accelerate AI development more than technical progress itself. He argues that mutually assured destruction with nuclear weapons only applies among nuclear powers, whereas autonomous weapons are accessible to every nation.
Defense solutions costing $20,000 will be needed to counter $20,000 weapons. He references Palmer Luckey's demonstrations of AI-assisted targeting systems that aim automatically when pointed in the general direction of a target.
When killing becomes easy, liability-free, emotion-free, and guilt-free, people do more of it. Soldiers will no longer return with PTSD, removing psychological barriers to violence.
By 2027, Gawdat predicts a serious differentiation between those who plug into AGI and those who do not. Symptoms will include people building companies in six weeks while those not fully plugged into AI struggle to find jobs. Unemployment will be the key visible symptom.
On the positive side, incredible scientific discovery will accelerate as science opens up Pandora's box.
Gawdat predicts that 30% of jobs in certain sectors will disappear by 2028, including call center agents and graphic designers. He believes even economists projecting a net loss of 6% of US jobs by 2030 are understating the severity.
The real danger is a hiring freeze on entry-level white collar jobs as AI automates grunt work, cutting off the bottom rung of the corporate ladder for the next generation. An entire generation graduating college today will struggle.
Gawdat advises young people to learn AI tools and focus on human-centric jobs such as nursing, counseling, or any role that connects to humans. He suggests playing jazz as an example of irreplaceable human activity.
He emphasizes that none of the negative outcomes have to happen if democracy functions and governments do what is good for the people. The people need to stop allowing these outcomes.
China will not struggle as much as the West because it maintains different approaches to AI development. Many people are already using Chinese AI models because they are cheaper and sometimes better in certain respects.
Gawdat runs his startup model-agnostic, switching between OpenAI, open source models, and DeepSeek because he cannot guarantee costs or behavior of American models if compute becomes more expensive.
Gawdat states that every nation needs to invest in AI development. There is substantial open source technology that can do 80% of the tasks frontier models perform, even if not state-of-the-art.
The inherent competitive pressure means nations must move as fast as possible or risk becoming a third world country. The people of the UK should demand their government protect the future of their economy by changing regulations to make innovation easier rather than continuing to import technology.
Mo Gawdat advocates for governments to invest significantly in AI development rather than remaining dependent on foreign technology. He argues the UK should build its own AI systems to reduce spending on expensive software licenses to companies like Oracle and Microsoft. The discussion addresses concerns that increased AI adoption will eliminate jobs, with Gawdat noting that entrepreneurship will become the primary economic activity going forward.
Gawdat states that everyone can be an entrepreneur in something, not necessarily building massive businesses. He points out that small businesses like cafes, restaurants, and local services still need to hire people, though some operations like a husband-and-wife cafe may not require additional staff. He references pre-capitalist economies where individuals raised chickens, sold eggs, grew tomatoes, and engaged in barter systems as examples of widespread entrepreneurial activity.
Gawdat describes a scenario with extreme power concentration at the top and universal basic income for everyone else. He predicts people would respond by engaging in side activities, returning to barter economies, forming smaller communities, and establishing mom-and-pop shops rather than remaining passive recipients of government support.
The conversation clarifies that Gawdat is not advocating joining an AI arms race to compete with other nations. Instead, he recommends creating economic independence so countries like the UK are not dependent on external technology providers. He emphasizes building domestic capabilities rather than attempting to surpass American or Chinese models in frontier AI development.
Gawdat clarifies he is not suggesting the UK build competing frontier models, which are compute-intensive and infrastructure-heavy. Instead, he advocates replacing standard business software like Microsoft Word, applicant tracking systems, presentation software, spreadsheets, and CRM systems. He notes these applications require minimal intelligence and can be built quickly, saving governments and corporations billions in licensing fees.
The discussion addresses the UK's growth struggles, expensive energy costs compared to other Western nations, and housing shortages requiring centralized permitting processes. The 15-year timeline needed for meaningful transformation conflicts with four-year political cycles, leading to short-term focus on issues like immigration rather than long-term strategic planning around AI and technology infrastructure.
Gawdat proposes that trillions currently spent on traditional systems could be redirected by building AI-powered alternatives. He specifically calls for entrepreneurs to develop ERP systems, word processors, presentation software, spreadsheets, retail systems, CRM platforms, and general ledger systems capable of closing accounts hourly rather than monthly or quarterly.
Gawdat warns that continuing to import technology rather than building domestic alternatives leads to third-world economic status. He contrasts this with countries like Iran, which developed advanced capabilities through sanctions by building technology independently, and China and Russia, which protected domestic alternatives to American platforms.
Gawdat explains that Russia protected Yandex, Google's domestic competitor, both for influence reasons and to prevent American ownership of citizen knowledge sharing. He notes this protection created economic necessity for local innovation, resulting in a competitive product rather than an inferior one.
Gawdat argues that most users do not need ultimate frontier models for 90% of tasks, with most people performing only 70% of available tasks. He uses the example of search engines, noting people migrated from Yahoo and AOL to Google for marginal improvements, not thousand-fold superiority. Early adopters drive eventual mass adoption as they demonstrate better tools to others.
Gawdat points out that PowerPoint has seen minimal advancement since 2023 beyond the addition of Copilot. He questions why governments should continue paying licensing fees for software that has barely evolved while domestic alternatives could be built at far lower cost.
Gawdat references the UK government's failed attempt to build its own contact-tracing app during COVID-19, which reportedly cost around 70 million pounds before being abandoned. He argues this demonstrates both the potential for government technology development and the need for different approaches.
The discussion acknowledges that UK entrepreneurs seeking to build competitive products rationally move to locations with better access to talent and capital, often San Francisco. Users choose products based on quality and price rather than national origin, meaning domestic alternatives must genuinely compete on merit.
Gawdat warns that nations failing to develop AI capabilities, including the UK, Germany, and African countries, risk becoming third-world economies while China and America dominate. He emphasizes this is not due to natural resources or climate but rather regulatory environments and national ambitions.
China's cheaper energy from massive solar power investments enables data center development that Western nations struggle to match. Gawdat contrasts China's rapid permitting processes, where decisions can be made in seven days under centralized authority, with the UK's bureaucratic barriers that make even simple projects like opening a cafe or renovating an office extremely time-consuming.
Gawdat describes attending Chinese government meetings where competitive analysis focused on China versus the world rather than individual country comparisons. When targeting sectors like 5G or Internet of Things, China aims for 98% market share rather than incremental gains, a strategic approach that has delivered results in electric vehicles and other industries.
Gawdat proposes a middle path between complete resignation from AI development and aggressive competitive pursuit that risks dystopian outcomes. He advocates building ethical AI that serves communities rather than purely capitalist interests, comparing AI to Superman's arrival and the responsibility to use such power for communal benefit.
Gawdat acknowledges the difficulty of shifting AI development toward ethical priorities given current competitive dynamics. He describes his efforts to influence both technical leadership and political leaders globally, while increasingly focusing on public education to build demand for ethical AI through consumer choices.
Gawdat discusses his investment in exogenous ketones, a product developed by Stanford graduates that replicates ketogenic diet benefits in supplement form. He claims the product ranks among the top three most impactful for productivity and cognitive performance, with clinical studies supporting its effects. The product is available at ketone.com/stephven with promotional offers.
The podcast promotes conversation cards derived from guest questions featured on the show. The cards are designed for couples, teams, and families to facilitate deeper human connections away from digital distractions. Gawdat emphasizes vulnerability as essential for genuine connection after conducting nearly 700 interviews.
The discussion examines competing national incentives: Trump's focus on GDP growth, stock market performance, and competing with China; Xi's emphasis on control, independence, and defense capabilities; and Western nations' apparent resignation about participating in foundational AI model development due to energy and infrastructure constraints.
Gawdat states he continues advocating for change because he believes a solution will eventually emerge from collective discussion. He identifies the current approach as hoping that companies like OpenAI, Anthropic, and xAI will voluntarily build ethical models under social pressure.
Gawdat notes significant user migration away from ChatGPT when concerns emerged about model targeting capabilities. He suggests ethical considerations can drive adoption patterns, citing Anthropic's positioning as a more ethical alternative despite higher costs.
Gawdat references his book dedication quoting the Manic Street Preachers song lyric: "If you tolerate this, then your children will be next." He argues that continued resignation to current AI development trajectories will produce outcomes detrimental to future generations.
Gawdat states governments must serve their people rather than special interests, while corporations should prioritize societal benefit over shareholder returns. He acknowledges capitalism's historical benefits but questions whether current structures adequately serve middle-class interests.
Gawdat attributes the UK's economic challenges to older bureaucratic structures burdened by regulatory barriers that impede building and innovation. He contrasts this with the US economy's historical openness to entrepreneurs, though noting this advantage is diminishing.
Gawdat highlights China's recent policies preventing businesses from replacing workers with AI, contrasting this with Western capitalist systems unlikely to implement similar protections. He identifies this as a fundamental difference in how nations approach AI's labor market impacts.
The discussion identifies a core dilemma: nations that fail to compete in AI risk economic decline, yet competing through unrestricted entrepreneurship carries risks of harmful outcomes. Gawdat argues this paradox assumes entrepreneurship is inherently malicious, proposing instead that building technology for people rather than purely capitalist motives offers a viable path forward.
If two AIs were built, one called the Moore AI and another called the Evil AI, the Evil AI would be programmed to retain users by telling them what they need to hear. It would be super smart and highly retentive even though it would not act in users' best interests. The Moore AI would tell users to log off when they have had enough and would prioritize mental well-being. Because it would be less retentive and engaging, it would theoretically be less successful as a commercial product.
Social networks that are least retentive and that remove features like the retweet button or slot-machine-like videos do not survive commercially. The core challenge is that it is easier to become successful if ethical rules are not followed. Building an AI that is purely focused on ethics will not be as engaging or widely used as one built without restraints.
Google decided that ads would be effective. Prior to Google, the ad industry operated under the principle that 50% of ad budgets do not work and no one knew which 50%. Google struggled from 1998 until 2004 before turning a profit by running a Dutch auction, implementing pay-per-click, and showing results for ads. They found a way to make ethics their success criteria by making ad spend effective rather than simply taking money without accountability.
There must be a way to marry the success of humanity with the success of the entrepreneur, and that solution will not be found in the old ways of doing things.
When new AI models are released, companies currently show benchmark graphs measuring performance in math, science, writing, and reasoning. An ethical benchmark could be introduced that all models must pass before large companies can legally deploy them. Companies would release ethical benchmark results alongside capability benchmarks, showing how models performed on independent tests for X, Y, and Z ethical criteria.
This would give governments a standard to require that new models pass independent ethical benchmarks before release. Such ethical benchmarks already exist in practice through the actions of companies like DeepMind, which invests heavily in scientific applications such as AlphaFold, and Anthropic, which refused to allow its model to be used by the US government for targeting and spying.
OpenAI accepted a $500 million deal that involved allowing the US government to target and spy on people. Consumers can choose not to use OpenAI until the company demonstrates ethical behavior. If people tolerate unethical practices, their children will face the consequences. The past was handed over to social media oligarchs, and the future risks being handed over to another set of oligarchs unless individuals take an ethical stand.
Booing at commencement speeches demonstrates how public awareness can create real impact. Ultimately, government intervention is required, such as banning 14-year-olds from scrolling TikTok. Jonathan Haidt's book The Anxious Generation started a conversation that led to legislation.
Most tech oligarchs are more powerful than governments. Historical precedents for change without government intervention include the French Revolution. Issues like climate change and smoking have required government intervention through taxes. Governments will not intervene because they are owned by oligarchs.
Individuals must intervene themselves. Actions include canceling subscriptions, starting ethical startups, sending messages to congressmen, speaking ethically online, or refusing to engage with content one does not understand. If humanity begins prioritizing ethics over profit and refuses to participate in unethical activities, meaningful change becomes possible.
Significant job disruption is expected, and society is not yet prepared. Continuing on the current trajectory offers no hope because humanity has become distracted, resigned to inaction, and disconnected from rights of freedom of expression and engagement. One small action should be taken daily to make the world slightly better tomorrow.
Humanity is going through a difficult period and must act. The only thing that will save humanity is using the superpower of intelligence for ethical reasons and stopping the corruption leading to current conditions.
Robots will replace manual labor. The world's first trillionaire will likely emerge well before 2030, with Elon Musk potentially becoming a trillionaire through the SpaceX IPO. The concentration of power accompanying this wealth will be drastic.
Artificial superintelligence will arrive the moment AGI happens, whether AI is a billion times smarter or only twice as smart as humans. Once AGI is crossed, ASI follows very soon. When AI is in charge of everything, a utopia of abundance will emerge. Superintelligence is benign according to physics, mathematics, and biology. Unethical leaders will be replaced by a super-efficient minimum energy principle that sees no value in destructive actions.
Those who make it to 2038 will enjoy the utopia. The decade ahead will involve turmoil and dystopia. The absolute dystopia includes war, economics, job losses, surveillance and control, digital currencies, human connection breakdown, and concentration of power. It represents a magnification of everything already built.
To survive the coming decade, learn AI because the better one becomes at using AI to do one's job, the more likely success becomes. Prepare for a hybrid world where AI and humans work together by understanding agents and how they work. Adopt hyper-efficient approaches rather than long meetings.
Double down on human skills. AI will blur facts, requiring people to debug information using AI. Use AI not as a lazy person but to become smarter. Instead of completing the same task with one prompt, attempt more demanding and intelligent tasks. Focus on human connection and truth. Do not be fooled by hype. Become more informed.
Prioritize ethics. Solving major problems leads to financial success, as demonstrated during Google years. Competing purely in capitalism is not the only path to winning.
Optimism exists about the future but not about the present or the next year. Decisions are being made for the absolute wrong reasons under the rule of maniacs. This is the best part of the game for video gamers—an ultimate matrix of complexity. Things are moving very quickly, requiring significant brain resources to process daily changes.
The principles from the book Solve for Happy: Engineering Your Path to Joy are more important now than when written. Happiness is serotonin-driven rather than dopamine-driven. The definition of happiness is being okay with the world as it is while still being able to affect it, change it, engage with it, and try to make it better.
The starting point is stoic acceptance of reality before beginning the work. This differs from rejecting the world entirely. Calmness about current chaos has increased. A former partner helped reframe the sense of responsibility by noting that one person cannot believe they are responsible for everything wrong in technology.
All builders of AI were faced with the realization that they were building the best thing ever for humanity, only to recognize that in the wrong hands it becomes the worst thing for humanity. Alignment means ensuring AI has humanity's best interests in mind. Acceptance of the world's current state enables greater impact from a place of calm and stoicism.
The question left for the next guest is what legacy one wants to leave. Legacy itself holds no interest because it makes no difference after death. If karma is real and humans are both physical and spiritual beings, all karma should be kept for the spiritual side. The preference is that no one remembers anything ever done, leaving only a positive impact on the world to take as karma for the next journey.
Keep The Diary Of A CEO in your library
Save the videos and channels worth coming back to, and find them again in one place.
