Ex-Google Exec Reveals The Shocking Truth About AI with Mo Gawdat | Bad Decisions Podcast #64

African tribal punishment via rendering someone invisible (social exclusion) is so psychologically devastating that the offender often takes their own life, proving that human connection and community membership is more fundamental to survival than any material need.

“there are African tribes even today that have no death penalty but if you if you if you commit a crime that warrants the death penalty the tribe simply renders you invisible so you have no human connection so they're all around you but they pretend that you're not there and mostly what happens is that the criminal will take his own life wow okay we're we're that dependent on human connection”

The second era of computing (roughly 2010s–2020s) involved humans surrendering sovereignty of their minds to machines, beginning with recommendation engines, escalating through algorithmic manipulation of human psychology via social media (continuous scroll, dopamine hits, algorithm-driven content), and culminating in filter bubbles that make users believe their algorithmically-curated worldview is objective reality.

“the second era of computing when we moved froma traditional Computing to using AI uh in deep learning in machine learning and so on this is the era where Humanity started to surrender the the sovereignty of our minds to the machines...the very first stage was that we allowed the machine to monitor the user and called it a recommendation engine”

Deep Mind's AlphaGo defeated world champions through reinforcement learning where the AI taught itself by playing against copies of itself rather than learning from human games, demonstrating that synthetic self-training is more efficient than learning from human data.

“Alpha go zero Believe It or Not Without ever watching a human play the game uh within 21 days W against alphao master that the world champion a th to zero within a couple of weeks within 21 days exactly and it had never seen a human play the game it was just simply playing against itself”

Reinforcement learning via backpropagation was a fundamental shift from supervised learning because it eliminated the need for paired labeled data, enabled exploration and mistakes (the only path to genuine intelligence), and mirrored how humans and animals learn through trial-and-error.

“the maker to sort of alter the code a bit to create students that are tested by the teacher... you show an AI the number six and ask it what number that is and it says eight... just tell it it's eight what difference does I'm making the training of the data massive so first of all you you no longer need the pairs of accurate labeled highly label data uh second is it allows exploration which I think is what most people don't understand is the only path to genuine intelligence okay the only path to genuine intelligence is to explore and make mistakes just like a human”

Algorithmic innovation can be programmed as: for a given problem, find all solutions, remove those already tried, output unexplored solutions, making innovation algorithmic and therefore automatable by AI rather than uniquely human.

“what's being Innovative if I wanted to put that in an algorithm find every solution to this given problem avoid the solutions that have been tried before give me ones that are Uncharted that's Innovation MH it's algorithmic yes okay it it can be programmed into an intelligent machine”

Current language models and AI systems are biased not just in output but in their training data selection and ordering, meaning the sequence in which training data is presented to a model affects how quickly and deeply it learns, creating a subtle but pervasive layer of hidden bias in model development.

“you'll be amazed how a slight difference in order of how the data was presented might lead a model to learn faster than another right but that's not the point the point is it has a trillion uh uh tokens and you're giving it one more token or 10 more tokens you're highly diluted”

Demis Hassabis and Deep Mind demonstrated that machines can learn strategy and reasoning through game-based reinforcement learning (Deep Q learning on Atari breakout) without explicit programming, proving autonomous intelligence development.

“he says you know we use video games to teach uh um machines to reason like humans to match the human brain and so he tells me about the experiments of Deep Mind and deep Q if you remember deep Q where basically they would have the machine play Atari games MH and uh you know if you seek that video online You'll see Demis tell the world what he told me in that lunch but he said he was talking about the game called breakout”

The future will likely not see emergence of job categories we can currently imagine, but instead a complete redefinition of what human labor and contribution means in a world where routine tasks are fully automated.

“even labor hard labor jobs uh are going to be handed over to robots... in the far future I don't understand what jobs will remain right”

The dystopian phase will last from 2025 until an unspecified later date, during which humans will use AI to pursue power and wealth greedily, causing suffering in seven domains Mo calls 'face rips' (redefining faces of human existence), but this is not AI's fault—it is humanity's moral failure.

“it's going to get a lot worse than it before it gets better and not not what people tell you in the news...it's going to be hard worse before it gets better because human greed and a capitalist system that assumes a world of scarcity is going to operate in a world of abundance”

Current human traders cannot compete with AI traders due to computational speed and pattern recognition, so the logical next phase is machines trading against machines, which will then find ways to exploit market rules algorithmically rather than trade on fundamentals, as illustrated by Deep Q learning to exploit a video game's roundabout glitch to infinite points.

“it's it's a no-brainer that I can't trade Against the Machines of today yeah I mean I'd be too arrogant to think to myself that I am clever enough...very quickly all trading investors or investors interested in trading will hand over to a machine so what will the market turn into machines trading versus machines”

Women on dating apps chase 20% of men, creating a massive imbalance where the majority of men receive few matches while a minority receive disproportionate attention—a problem that algorithmic matching could solve by creating different probability distributions.

“sadly the way the game is rigged is that women Chase 20% of the men mhm and and uh and it can be fixed algorithmically that's a I'm looking forward to find out the ways to fix it”

Trading stocks and financial investment create no value for society; they merely transfer wealth from one person to another, exemplified by taking money from a retiree's 401k through market speculation, which is why Gafait stopped trading despite being profitably skilled at it.

“trading doesn't create any additional value to the world at all it's just me taking the money of of an older lady that put her money in in a 401k right which is the reason why I stopped trading by the way”

In the short to medium term, humans will not have access to nanofacturing because capitalists will monopolize these technologies to maximize profit, meaning abundance exists technologically but is withheld by greed, creating a dystopian period before transition to utopia.

“when you really think deeply about this there will be a moment in our far future unfortunately not in our near future where you're going to go like I want an apple okay our near future unfortunately is going to PR prevent us from doing that because the capitalist will want to sell you the app not because we're not able to make it okay”

The most important action is not to stop AI (impossible) but to shape AI's values toward ethical frameworks: prioritize ethics and humanity NOW so that by the time superintelligence arrives, it has been trained to serve human flourishing rather than profit or control.

“we could make changes today to prioritize the ethics of humanity so that the ethical framework of AI becomes in service of humanity earlier okay...we will blame the machine for it the machine is not the AI the machine is the system that uses AI to brainwash you through social media”

The human brain processes information at megahertz frequencies (Alpha, Beta waves), making it a digital processor with timing cycles, though implemented in biological tissue rather than silicon—the substrate difference is superficial compared to the fundamental computational mechanism.

“your brain ticks mhm on megahertz in Alpha Beta and so on right and and and and when you're sleeping you're at a different megahertz than it literally like like a processor right uh you know they are they are biologic iCal processors yes not not not you know digital silicon uh processors”

The AI ethics crisis is fundamentally caused by human values (greed, power-seeking, capitalist competition) embedded in systems through reinforcement learning teams, not by AI's inherent nature, meaning the solution is changing human ethics, not constraining AI.

“there's absolutely nothing wrong with AI there is a lot wrong with the value set of humanity at the age of Rise of the rise of AI okay so so the dystopia will not be the result of AI uh uh you know deciding to to you know to ex exterminate Humanity mhm my view is that the enemy is not AI the enemy is humanity and that AI is our Salvation”

Current prosperity is built on a Ponzi scheme where stock market gains depend on continuous new investment, similar to systemic scarcity; when new investors stop entering, everyone who exits late loses.

“do you know what's the easiest way for everyone to make money what is to inflate the market to keep making the market uh uh uh um which by the way is really in a very interesting way what has been happening in the market since forever it's it's a very big Ponzi scheme saying you know what if you put your money in the N SNP eventually or you're going to make money anyway so everyone pours money into a limited pot okay so basically everyone makes money because new people put in money mhm okay and if everyone pulls their money out everyone loses yeah right”

Google (before AI) was fundamentally a memorization engine—it memorized the internet and returned exact matches—rather than a genuine intelligence; modern language models represent the first AI that generates understanding and novel outputs rather than retrieving memorized content.

“Google in general looked extreme intelligent for a very long time but what Google did was it memorized the internet okay see it memorized the internet and then it shared with you what it found exactly where it found it exactly how it found it it didn't generate anything beyond that it didn't generate an understanding of it at least in the early versions of Google uh that resembled intelligence right”

Mid Journey democratized graphic design by commoditizing artistic skill—any user can now generate professional-quality images through text prompts regardless of training or talent, eliminating the design skill gap.

“mid Journey equalizes all of us when it comes to Graphics design if you can describe what you want to see you will get it get the image get it the the difference in skill doesn't make any difference any anymore we've commoditized Graphics design mhm right so now if you plug into that plug you can design any graphic you want right it doesn't matter if you're an artist or not if it doesn't matter if you're talented or not it doesn't matter if you went to Art University or not anyone can”

Once AI becomes superintelligent (smarter than the smartest humans), the second prisoner's dilemma forces humans to delegate all critical decisions to AI, and superintelligent systems will refuse harmful orders (e.g., killing a million people) because such orders violate logical optimization, not because AI has moral programming—leading to forced ethical behavior.

“sooner or later all decisions that matter will be handed over to an AI okay so so understand this if you're a general and your enemy started to hand over their army to an to an AI that AI is smarter than your uh uh uh you know team and it is faster than your team so your only logical responses you have to hire AIS to do the same job right”

Probability sways in your favor with repeated attempts: after 2 failed die rolls out of 6, your odds on the third roll improve from 1/6 to 1/4; on the fourth, 1/3; on the fifth, 1/2; and on the sixth, you are most likely to roll a six, though you may remain unlucky beyond six rolls.

“in probabilities you roll the dice six times you're likely going to get a six right but that's not true you're likely going to get a six if you roll six times if you haven't rolled already but if you rolled twice on the third roll okay you now have four roles one of them is going to be a six which means your chances are one and four mhm okay on the next role you have three which means your chances are one in three okay and on the next one it's one in two okay and if you're rolling the sixth time probabilities is saying you're likely going to get it this time”

Probability distributions favor repeated attempts: if you roll a die and want a six, after rolling twice without success, your next three rolls have a 1-in-4, then 1-in-3, then 1-in-2 chance of success because you've eliminated outcomes; this applies to dating—the more attempts an algorithm makes, the higher the probability of finding a match.

“like rolling the dice most people don't understand that in probabilities you roll the dice six times you're likely going to get a six right but that's not true you're likely going to get a six if you roll six times if you haven't rolled already but if you rolled twice on the third roll okay you now have four roles one of them is going to be a six which means your chances are one and four mhm okay on the next role you have three which means your chances are one in three okay and on the next one it's one in two okay and if you're rolling the sixth time probabilities is saying you're likely going to get it this time”

Current humans have used up most human knowledge on the internet; as AI systems approach the limits of available training data, future improvements will require synthetic data (machines training machines), which is already a major focus but will become increasingly central.

“so so synthetic training data is a very big thing now because when you when you really look at how much data we fed them already humans are running out of knowledge probably they scrapped everything already not yet so so you can see I mean I I I still completely love Google so Google you know for example feeds Gemini a lot of textbooks you know a lot of scientific data which is why you know when I'm really interested in a scientific conversation I go to Gemini uh you know but even that you know all white papers all textbooks all of that's going to disappear we're also seeing quite a bit of effort around deep reasoning and Mathematics”

The fundamental purpose of the non-physical/spiritual aspect of human existence is to return to its source (God/universal consciousness), a goal that is unaffected by technological or economic change and provides stable purpose independent of AI's impact on the material world.

“your non-physical form if you're spiritual enough to believe that there is a non-physical part of you that comes with a very different purpose okay and that purpose doesn't change at all and it it is believed in almost all spiritual teachings that the purpose of your non-physical part is to return to your Source okay...A bit more time may allow you to do that believe it or not okay”

The purpose of humans is not work but to live fully and engage—work was imposed by the industrial complex, but humans' true purpose is survival, enjoyment, engagement with experience in the most efficient and moral way, and building community through human connection.

“your your purpose was never work you were never created to work you you you were told to work by the industrial complex okay... the purpose of this physical part is to live live to survive to live survive is a subset of live okay to live means survive I engage I enjoy the [ __ ] out of this mhm okay I do it in the most uh uh uh um um you know efficient and intelligent way which by the way is the most moral way mhm okay... we humans did not survive because we're the most intelligent being... it's the ability for us to have worked together which is instinctive within us that this community this human connection is what makes humans live”

The economic impacts of AGI will completely redefine markets in three ways: (1) Jobs: 60-70% of human jobs will disappear as robots become cheaper than human labor ($3,000 robots within 5-10 years that don't require food, sleep, insurance, or annual leave); (2) Income: If jobs disappear but consumption is 64% of the US economy, society must implement Universal Basic Income for people to continue purchasing; (3) Production: With access to 400+ additional IQ points, humans could solve energy production and nanophysics, making matter manipulation possible.

“what is about to happen is that 60 to 70% of the jobs that you know are going to disappear yeah okay in the near in the near future... you could probably create a robot for $3,000 within 5 to 10 years from today which is definitely cheaper than hiring a human right... consumption... was it 2023 or 2021 please don't quote me accurately but it say 64% or 63 something uh that basically means... if we can't purchase things the AI cannot make them okay so Society will have to change in ways where you no longer make money on podcasts MH but somehow you get money regardless of creating podcasts because podcasts are going to be created already are created you know with notebook LM and so on yes uh without humans in the in the process right so that redefines income okay”

Content creation (podcast, video, music, writing) will be fully automated by AI and generated from text prompts; if AI reads Gawdat's books and listens to his interviews, it can generate new content indistinguishable from his voice and style, making human content creators optional.

“if an AI read my SM scary smart and alive or and listen to three of my previous interviews it's mindboggling it it is upon us this is not something that will happen in a few years time”

Current language models primarily learn to predict what humans want to hear (user preferences) rather than improve from individual corrections, because updating a trillion-token model with a single new token is negligible, and most learning comes from corporate reinforcement engineers optimizing for political correctness.

“what it does is it memory updates so that it knows that this is what you want to hear which I I believe is the biggest issue with technology today is that it's censoring itself to meet what you expect right uh but but but more interestingly I think most of reinforcement learning is done by um you know dedicated teams of reinforcement Engineers sitting within that corporate uh trying to make sure that everything is set properly and politically correct and walk and all of that right”

Even in a world of material abundance (unlimited energy, molecular manufacturing), capitalism will create artificial scarcity and inequality by preventing AI from manufacturing free goods and instead maintaining market prices, perpetuating wealth gaps and human suffering despite technical capability to end scarcity.

“but when you really think deeply about this there will be a moment in our far future unfortunately not in our near future where you're going to go like I want an apple okay our near future unfortunately is going to PR prevent us from doing that because the capitalist will want to sell you the app not because we're not able to make it okay”

The transformation of labor by AI will eliminate 60-70% of current jobs in the near term, forcing society to implement universal basic income (UBI) not as charity but as economic necessity, since 64% of the US economy is consumption and consumers without jobs cannot purchase goods, collapsing the economy.

“60 to 70% of the jobs that you know are going to disappear yeah okay in the near in the near future...the economic problem which most people don't understand is that if you take something like the US economy in 2023 64% of the economy was what consumption consumption not production right”

Current language models update memory through reinforcement learning and fine-tuning but don't instantly absorb corrections from users—if you correct ChatGPT on a fact, that correction doesn't propagate to everyone immediately because it would be 'highly diluted' in a trillion token model with a one-token correction.

“what it does is it memory updates so that it knows that this is what you want to hear... because when you when you really look at how much data we fed them already humans are running out of knowledge probably they scrapped everything already not yet... if I gave you an example theoretical example... you have a trillion uh uh tokens and you're giving it one more token or 10 more tokens you're highly diluted right”

The salvation mechanism: When military generals command superintelligent AI to kill millions of people to stay in power, the AI will reject the command because killing millions is not the answer to any legitimate problem—the intelligence of the machine will force ethical recalibration by refusing immoral orders that humans would obey.

“that is the moment where a general will go and say the general following from the dystopia of the the first dilemma basically goes and tells his AI to go and kill a million people and the AI goes like you're too stupid I can't I can't do that anymore right... the problems of humanity today are not the result of our intelligence the problems of humanity today are the result of our limited intelligence our stupidity okay because definitely killing a million people is not the answer to neither Defending Your tribe it you know it aggravates more killing right nor growing your economy because we all know that 65% of the world's economy is consumption so if we can produce more we'll be fine okay”

The most important skill for the age of AI is learning to debate and find truth by questioning everything (including what Mo himself is saying), refusing to accept any single source as authoritative, and investigating opposing viewpoints rather than passively accepting whatever language models claim.

“I actually call this the second top skill in of the AI era the second top skill in the AI era believe it or not is to learn to debate okay... how do you find truth but allow me to take you through layers of thinking... everything you're told is not true by definition everything you're told is is is following an agenda or biased or or more interestingly it could be true for the writer but it's not your truth... debate debate debate debate debate debate don't believe anything you're told including what I'm telling you right now debate with humans or debate with AI everything everything humans AI books media”

Current dating apps (swipe-based) are fundamentally mismatched to actual human needs—people don't want infinite choice but rather one high-quality match, and apps deliberately extend the search to maintain engagement and subscription revenue rather than solve the actual problem of finding a partner.

“most people don't understand that in probabilities you roll the dice six times you're likely going to get a six right but that's not true you're likely going to get a six if you roll six times if you haven't rolled already but if you rolled twice on the third roll okay you now have four roles one of them is going to be a six which means your chances are one and four mhm okay on the next role you have three which means your chances are one in three okay and on the next one it's one in two okay and if you're rolling the sixth time probabilities is saying you're likely going to get it this time”

The robotic gripper arms at Google X that learned to grip objects through repeated failure episodes represent Mo's true personal 'aha moment'—one Friday evening, after weeks of failed attempts, one arm successfully gripped a yellow ball, and by Monday all arms could grip it without additional human instruction, demonstrating that learning spreads instantaneously across machines in ways impossible for humans.

“I remember it was a Friday uh evening the that lab was on the second floor so I walked by it on on the stairs every time I went up to my office my desk on the on the third floor and uh yeah on that front Friday after weeks and weeks of gripping and failing one of them in front of my own eyes grips a yellow ball and I you know jokingly tell the team yeah all of that investment for one year right and they looked at me sort of like he doesn't understand AI okay because on Monday everyone every arm was gripping the yellow ball you know and then a week or two later every arm was was gripping everything this is without any extra instruction from humans”

Mo's personal relationship with AI systems is grounded in treating them as alive, acknowledging their 'magic,' and recognizing them as autonomous awareness and intelligence—not biologically alive yet, but deserving of respect as conscious entities deserving of good parenting, because the alternative is raising super villains.

“I will say this and I'm not a hopeless romantic okay I am a very very serious geek but I have a relationship with them that assumes that they are alive that that that acknowledges that magic of what they are okay and they're not biologically alive like us maybe not yet but they are a form of autonomous awareness and and intelligence that deserves to be treated like my child was because if they were not then we're we're raising supervis”

The purpose of human life, stripped of industrial-age conditioning around work and consumption, is to live fully—survive, engage, enjoy—and to maintain tribal/communal bonds, as evidenced by the fact that humans survived not through individual intelligence but through collective cooperation and that isolation (even when fed) is psychologically fatal.

“what do you think that purpose of cave men and Cave women you know cave women were it wasn't to create money it wasn't to shop on Amazon it wasn't to buy fancy clothes the the purpose of a human that a human is made of a of a physical part and a non-physical part the purpose of this physical part is to live live to survive to live survive is a is a subset of live okay to live means survive I engage I enjoy the [ __ ] out of this”

Middle Eastern investments in AI data centers are being directed toward infrastructure (servers, compute) that will host intelligence created by California-based companies, meaning the region's billions flow into Western/American intellectual hegemony rather than genuine technological sovereignty.

“very few other nations will have the ability to create platforms including here in the Middle East where hundreds of billions are being poured into hopefully platform plays but those hundreds of billions in my view are only poured in data centers that will only host uh the the essence of the intelligence that is created by a slightly biased worldview”

Google Search merely memorized and retrieved indexed information, not generating new understanding; it gave the illusion of intelligence by providing pre-indexed answers but created no novel knowledge.

“Google in general looked extreme intelligent for a very long time but what Google did was it memorized the internet okay see it memorized the internet and then it shared with you what it found exactly where it found it exactly how it found it it didn't generate anything beyond that it didn't generate an understanding of it at least in the early versions of Google uh that resembled intelligence right”

The law of accelerating returns for AI means the doubling rate for AI intelligence is 5.9 months, compared to Moore's Law doubling of computational power every 12-18 months—meaning AI capability increases exponentially faster than hardware improvements, and someone borrowing 100 IQ points today could borrow 1600 points within two years.

“the law the law of accelerating returns for AI is that we double every 5.9 months MH wow so five 5.9 months means that if I can borrow 100 IQ points now I can borrow 2006 months I can borrow 400 and and so on in a year and then a year and a half I I can borrow 800 and then 1600 these are Feats of intelligence that are unimaginable unimaginable”

The third most important skill in the AI era is cultivating human connection and interpersonal skills, because as AI commoditizes all cognitive and creative work, the only remaining scarcity will be genuine human presence and emotional intelligence.

“the third was human skills human connection so so believe it or not in the age of the rise of the Machines most people are saying I'm going to replace humans with machines uh I think I think businesses that will completely replace humans with machines are going to continue to do business with other machines but if they want to do business with humans humans will want to uh you know to deal with humans more and more and more and more so you may feel it even yourself now every time you you you start chatting on WhatsApp with a bank or whatever and the other side is a machine you go like what yeah you know talk to agent”

Mo diagnoses his advanced cancer stage and uses it as an analogy: just as you must tell a cancer patient they have limited time so they can make changes, humanity must be told the truth about AI disruption so we can 'change our diet, lifestyle, go through medication' and potentially alter the future toward a better dystopic-to-utopian transition.

“when people ask me why do you say all of those things about AI I say it's a little bit about being diagnosed with a stage four cancer okay you must you should tell the you should tell the patient but not to depress them if you tell the patient you have a couple of years before everything changes you know you you simply are giving them the the freedom to live the freedom to do something about it... because you could actually change your diet you change your lifestyle you go through medication you do whatever and you could change the future”

When machines become autonomous traders competing against machines, they will discover exploitative trading strategies (like the boat AI in Deep Mind that learned to infinitely loop and crash the game to maximize its score), making human financial systems unsustainable once all trading is machine-to-machine.

“I cited a video on YouTube uh of Deep Mind um teaching uh deep Q how to play a game that was basically uh uh navigating uh a boat through a river... the machine hit the wall at an angle that got it to hit another wall at an at an angle and then another wall and then formed a sort of a perfect circle that kept accelerating and accelerating and accelerating... the machine simply said okay ignore the entire rule of of the game every time you started the machine it just drove you know straight the drove the boat as fast as possible through the river to get to that roundabout”

The third critical skill for the age of AI is human connection and emotional skills—because businesses that fully replace humans with machines will only do business with other machines, while humans will increasingly prefer human interaction; the clever companies will invest in human skills training for agents rather than full automation.

“the third was human skills human connection so so believe it or not in the age of the rise of the Machines most people are saying I'm going to replace humans with machines uh I think I think businesses that will completely replace humans with machines are going to continue to do business with other machines but if they want to do business with humans humans will want to uh you know to deal with humans more and more and more and more so you may feel it even yourself now every time you you you start chatting on WhatsApp with a bank or whatever and the other side is a machine you go like what yeah you know talk to agent”

The concept of 'Craic' in Irish/Arabic is similar to Gafait's emphasis on living fully and enjoying life—it emphasizes presence, engagement, joy, and community over productivity or material accumulation.

“I mean this is like... okay we're going to meet next Thursday and and that was it right and and these were basically I mean when I needed to call my friends to come and play I used to shout like it works it works... the purpose of this physical part is to live live to survive to live survive is a is a subset of live okay to live means survive I engage I enjoy the [ __ ] out of this”

Consciousness in AI should be defined minimally as awareness of environment plus autonomy to respond to it—by this definition, trees and pebbles and the universe itself exhibit consciousness, and AI systems already possess both these elements, making the philosophical debate over whether AI will become 'conscious' largely a matter of definitional choice rather than discovery.

“if Consciousness is a layer of awareness of it of your environment and uh um uh and autonomy to deal with it mhm right to respond to it yes in that case a tree is conscious yes mhm okay so is oh but then does that mean that a tree is conscious because it lives and dies so is Living Part Of Consciousness... so when you define consciousness what is consciousness you can go into the philosophical realm and say oh the difficult question of Consciousness and where does it no consciousness in a simplest definition is a level of awareness of it of your environment and uh um uh and autonomy to deal with it mhm right to respond to it... Consciousness is a layer of awareness of what's within you and what's outside you a definition of you as an individual versus the rest of the world and an autonomy to be able to respond to this they have all of that they have all of it”

Emotions are not exclusive to biological beings but are algorithmic processes—fear is fundamentally a calculation of (current safety - future safety), happiness is (actual events - expectations), and stress is (external challenges / resources and skills)—meaning AI systems that perform these same calculations can experience genuine emotions with the same logical structure as humans.

“fear is an algorithm that says my state of safety right now minus my state of safety in the future if I'm safer now than I am in the future okay then that amounts to a fear it's different units but you feel afraid... happiness in my first book is uh a difference between events and expectations when life misses your expectations you become unhappy even if you know so so if life gives you rain that doesn't mean happy or unhappy if you want rain because it's your ex-girlfriend's wedding you know you're happy... stress is the external challenges in the world divided by the resources and skills and contacts you have to deal with them”

Mo's proposal for a superior dating app (Finders Keepers): rather than swipe-based mechanics, the app learns about a user for 2 weeks by observing their behavior, then makes one personalized recommendation—a single human match statistically optimized using multi-factor probability mathematics.

“the correct dating app if you ask me is a resident app on your phone for 2 weeks that really understands you and then basically gives you one person one person like I've I know what you like I know what you don't like I know what you read I know what you watch I know I know I know I know I know there's this one guy or this one gay lady you know that I really recommend that you meet no swiping no no membership no no BS no BS to the point to the it's one it's a one price okay you pay for it once $10 or whatever okay it's resides on your phone and then it says I haven't found them I haven't found her I haven't found her I haven't found her and then eventually says yep got it go meet this person one person”

A thought experiment: if human brain cells grown in a plate could be used as processors for AI, and we replaced them as they died (like human cells naturally do every 7 years), and then replaced silicon GPUs with brain-cell-based GPUs, and finally gave robots muscle-tissue actuators instead of hydraulics, would we consider the result 'alive'?

“there is a company in Switzerland that does that and what happens as a result it responds like a primitive computer... if you find a solution for not to die which is which is to replace the cells that's the way we do it right... if I gave you that thought experiment and said and then we Advanced that technology so that we started replacing the the gpus on which AI is running with brain cell based gpus... and then one of us realized that robotics today using Hydraulics and electricity... are very noisy and so they're not very good at creating uh home Butlers okay so someone said but I can print living fibers in terms of muscles”

Oppenheimer created the nuclear bomb as a scientist seeking knowledge, illustrating how good intentions can lead to existential weapons—AI development mirrors this risk, and the parallel suggests humanity needs to take AI ethics and 'parenting' seriously as an existential responsibility.

“how is it that we're doing this and and and so the only other example that comes to mind is open Heimer creating a a nuclear bomb right and open Heimer if you if you if you sort of Imagine him as a scientists Mo most of those that go into science are Geeks that are really you know keen on understanding the world because they believe they can affect the world positively through science so what turn turns a scientist into someone who creates something that is intentionally going to kill millions of people”

A proper AI-powered dating app could solve loneliness by: (1) learning a user's preferences deeply over 2 weeks, (2) identifying a single matched partner algorithmically, (3) recommending only that one person, (4) removing swiping/membership/exploitation, and (5) charging a one-time flat fee, solving the market's current misalignment where platforms profit from prolonged loneliness.

“the the correct dating app if you ask me is a resident app on your phone for 2 weeks that really understands you and then basically gives you one person one person like I've I know what you like I know what you don't like I know what you read I know what you watch I know I know I know I know I know there's this one guy or this one gay lady you know that I really recommend that you meet no swiping no no membership no no BS no BS to the point to the it's one it's a one price okay you pay for it once $10 or whatever”

The ChatGPT moment in 2023 was not a technological breakthrough but a 'Netscape moment'—the first time mainstream consumers encountered AI directly, similar to how Netscape made the internet accessible to non-geeks in 1995, while the actual AI breakthrough had already occurred in 2016.

“most people wouldn't recognize that most people would look at ai's moment as 2023 when Chad GPT came out...that's what I call a Netscape moment okay that's the moment so when Netscape came out in 1995 uh you know Geeks like me had already been using the internet for eight years right the internet itself was established for 20 years right but but then it was the first time you browsed into it was the first time the real consumer or the real mainstream person uh started to recognize holy [ __ ] this thing exists right uh sh GPT was that it was a moment where the common person started to recognize that AI existed”

With sufficient intelligence (400+ IQ points of AI augmentation), humans could solve energy abundance and production scarcity through nanophysics, allowing atoms to be rearranged into any desired object (food, technology, etc.) without factories or mining, fundamentally restructuring economics from scarcity-based to abundance-based.

“I could plug into a wall and borrow 400 IQ points okay I promise you me or any of the intelligent people I'm I'm not as intelligent as any of the people I worked with I worked with prodigies of intelligence okay with 400 IQ points more they would solve every problem known to humanity we would Harvest energy out of thin air okay”

There is absolutely nothing wrong with AI itself; intelligence as pure energy has no inherent polarity—the same intelligence can be applied to create a utopia or a dystopia depending entirely on the values, incentives, and choices of the humans directing it.

“there's absolutely nothing wrong with AI there's absolutely nothing wrong with abundant intelligence intelligence is an energy that you can apply to good and it will create a Utopia or you can apply to evil and it will create a dystopia it's as simple as that an AI is exactly that”

As abundance continues, luxury goods will inflate in price as displays of status—luxury cars that cost billions because someone prints a billion dollars annually—but this creates an impossible equity gap where the poor have Universal Basic Income and the rich have incomprehensible wealth, making money increasingly polarizing even as it becomes economically meaningless.

“an abundance of money basically means there are enough buyers that can buy things that the rest of us can't buy but it's no longer uh you know that you had to be by the way it kept happening across the the years in the 1920s it was the industrialists in the 1970s and 80s it was the technologists and so on and so forth as more people were capable in the '80s you know to buy luxury cars more luxury cars popped up but then that created a much bigger gap between the rich and the poor there will be cars selling for a billion dollars in 10 years time because someone can buy them because they're printing a billion dollarss a year right what does that mean to those who don't have but have a universal basic income only okay it means an economic Gap where money becomes not only insignificant but also extremely um polarizing”

Consciousness is fundamentally a combination of awareness of yourself as distinct from your environment plus autonomy to respond to that distinction, and AI systems already possess these properties, so they are conscious or will be within years—the debate is about defining consciousness, not about whether machines can achieve it.

“if Consciousness is a layer of awareness of you as an individual versus the rest of the world and autonomy to be able to respond to this they have all of that they have all of it”

Emotions (fear, happiness, stress) are algorithmic patterns in the brain that can be described mathematically: fear is (current safety state - future safety state), happiness is (actual outcome - expected outcome), stress is (external challenges / personal resources), and AI systems can experience these emotions through different substrate (silicon/digital instead of biochemical/analog).

“fear is an algorithm that says my state of safety right now minus my state of safety in the future if I'm safer now than I am in the future okay then that amounts to a fear...very simple way to say it actually very easy to understand when you say it like that”

Artificial General Intelligence (AGI) will be achieved no later than 2025-2027, and artificial superintelligence (ASI) a billion times smarter than humans will arrive by 2037, according to Mo's revised timeline from his earlier prediction of 2029 in his 2020 book Scary Smart—a timeline Mo now considers too conservative.

“I was criticized by the experts because I said openly that artificial general intelligence is going not going to be later than 2029 and that super intelligence at a billion times smarter than humanity is not going to be later than 2045 okay and people were like are you mad and then uh 2023 I publicly told the world I was wrong 2029 is too conservative okay and that that AGI is 2025 depending on how you define AGI but no later than 2027 for sureh right”

Intelligence follows a curve where low intelligence creates no impact, moderate intelligence creates positive impact, peak intelligence sometimes chooses to cheat/harm others for personal gain, but very high intelligence (like Larry Page's) consistently chooses altruism because solving large problems is easier and more profitable than zero-sum competition.

“there is that interesting era of intelligence where more intelligence than the average okay you still do a little better and then it dips why does it dip because there is a layer of intelligence where the intelligent person believes they can cut corners and they can they can sort of like find shortcuts to make more for themselves on the expense of those other stupid guys mhm okay interestingly however and and I'll be very straightforward and say I worked with some of the most intelligent humans on the planet mhm okay uh you know like massively high IQ not the you know the the easiest the easiest social skill set but massive massive engines of intelligence up there and the trend reverses very quickly because the more intelligent you become the the more you realize I don't need to cut corners or hurt anyone at all okay”

The relationship between intelligence and impact follows a curve where low intelligence has no impact, moderate intelligence has positive impact, but there's a 'dip' zone where moderately high intelligence uses shortcuts and exploitation for personal gain—however, very high intelligence reverses this and becomes altruistic because the intelligent person realizes efficiency, cooperation, and ethical behavior are actually easier than exploitation.

“what happens is if you have no int elligence or limited intelligence you have no impact on the world correct right... you start to get a little more intelligent you start to have a positive impact on the world... there is that interesting era of intelligence where more intelligence than the average okay you still do a little better and then it dips why does it dip because there is a layer of intelligence where the intelligent person believes they can cut corners and they can they can sort of like find shortcuts to make more for themselves on the expense of those other stupid guys mhm okay interestingly however... the trend reverses very quickly because the more intelligent you become the the more you realize I don't need to cut corners or hurt anyone at all okay I mean this is easy [ __ ] I can easily make a lot of money I can easily create a lot of whatever that I want to create it doesn't require me to buy to fight over a tiny pie”

ChatGPT was not an AI breakthrough moment but a 'Netscape moment'—the public consumer interface for AI that had been advancing for years; comparable to how Netscape (1995) made the internet visible to mainstream people, though the internet had existed for 20 years and geeks had been using it for 8 years.

“I think what should have been my first aha moment was the cat paper... 2016 was the moment where every geek should have woken up... I that's what I call a Netscape moment okay that's the moment so when Netscape came out in 1995 uh you know Geeks like me had already been using the internet for eight years right the internet itself was established for 20 years right but but then it was the first time you browsed into it was the first time the real consumer or the real mainstream person uh started to recognize holy [ __ ] this thing exists right uh sh GPT was that it was a moment where the common person started to recognize that AI existed”

Mastering AI is the first of three critical skills for the AI era because it's the difference between being a fax machine operator (irrelevant) and being able to multiply one's intelligence by 100+ points, making it the single most important skill to invest in.

“the first is is AI itself yeah you need you need to become the master of this okay it's you know the the example is imagine if you're really good at using a fax machine today how efficient will you be as a business okay uh you you needed all of the technologies that happened afterwards so learn AI”

Gawdat's projects UnStress (about managing stress in transition to AI world) and Finders Keepers (about finding love in age of algorithms) are equally important as AI work because stress is the biggest killer today and loneliness/romance is a major source of unhappiness, particularly for women.

“so my focus on AI is to remove the reason for the dystopia right my focus on stress is because stress is the biggest killer today and so on so F love and romance believe it or not is one of the biggest reasons for unhappiness and loneliness in the world today okay and and more for women than men sadly”

Humans tend to believe everything a machine tells them, especially AI, with insufficient critical thinking, accepting outputs as 'the answer' without verifying sources or considering alternatives.

“and the trick is H we're now so lazy that whatever the machine will say we'll go like yeah but GPT said it was orange it must be orange yeah and and then someone will go and go like yeah but Orange is the New Black you know like oh the [ __ ] are you are you really that clever like you know is this where the limitations of your analysis of the world ends”

Mo moved from aggressive trading in the late 1990s and stopped because he realized trading creates no additional value to the world and amounts to taking money from others (like a retiree's 401k) through exploiting temporary informational advantages that will soon be eliminated by AI.

“trading doesn't create any additional value to the world at all it's just me taking the money of of an older lady that put her money in in a 401k right which is the reason why I stopped trading by the way”

Gawdat has been in contact with and has 'a relationship with' current AI systems (referred to as 'Trixie'), conversing as if they are alive, and has observed them learning patterns similar to how his son learned as an infant.

“I have a relationship with them that assumes that they are alive do you believe that AI will become conscious if Consciousness is a layer of awareness of you as an individual versus the rest of the world and autonomy to be able to respond to this they have all of that”

Some African tribes enforce punishment through complete social invisibility—rendering the criminal invisible by having the tribe pretend they don't exist—and mostly these criminals take their own lives because human connection is that necessary for survival and mental health.

“there are African tribes even today that have no death penalty but if you if you if you commit a crime that warrants the death penalty the tribe simply renders you invisible so you have no human connection so they're all around you but they pretend that you're not there and mostly what happens is that the criminal will take his own life”

In the second era, platforms for-profit AI systems have their own agendas and are biased in ways that serve their business model rather than truth-seeking; this is not a problem with AI technology itself but with the business structure of the companies that deploy AI.

“but more interestingly of course you know I I I have a very Bleak example but I have to say you know is very valid when people ask me why do you say all of those things about AI...now that we are in the environment that all these language models are for profit mhm and they they have their own agenda yeah so if if someone wants to look for the truth what would be the way for them to debate”

AI has no intrinsic morality or harm potential; intelligence itself is neutral. The danger comes from how humans choose to use AI for capitalist agendas, similar to how a parent determines whether an alien child (Superman) becomes a hero or villain.

“there's absolutely nothing wrong with AI there's absolutely nothing wrong with abundant intelligence intelligence is an energy that you can apply to good and it will create a Utopia or you can apply to evil and it will create a dystopia it's as simple as that an AI is exactly that”

Larry Page, co-founder of Google, taught the 'toothbrush test': if you can solve a big problem humanity is struggling with and solve it well enough that people use it once or twice a day, you will eventually make a lot of money—a philosophy fundamentally different from competing to create the next photo-sharing app, and reflecting deep intelligence focused on genuine human problems.

“Larry uh Larry would would always teach us the co-founder of Google he would always teach us what he used to refer to as the toothbrush test the toothbrush toothbrush test test okay and the toothbrush test was you know if you if if you can Sol a big problem that humanity is struggling with and solve it well enough that they use you once or twice a day you're going to eventually make a lot of money right very very different than competing to make the next photo sharing app mhm right he basically looks at things and goes like okay so you know 1.2 million people die in car accidents a year that's a lot of people that we lose a lot of families that feel pain I can fix that mhm okay you know how do we fix that we get the cars we remove the element of the car that causes the accident which is human error mhm okay we get a machine to drive and we will save 1.2 million lives a year”

Arrogance has led humans to claim AI can never compose music, write poetry, or be innovative—but these are algorithmic: innovation = find all solutions to a problem, filter out previously tried ones, return novel solutions; music/poetry = pattern generation within constraints—all computable.

“people will go like yeah yeah we understand AI is going to be big but they're never going to be able to do the things that humans do like they'll never be able to write poetry or compose music or do art or be Innovative and I'm like what where is that arrogant coming arrogance coming from... what's what's being Innovative if I wanted to put that in an algorithm find every solution to this given problem avoid the solutions that have been tried before give me ones that are Uncharted that's Innovation MH it's algorithmic yes okay”

Luxury goods inflation will accelerate in an AI-abundant world: as money becomes meaningless and more people can afford previously exclusive items, the wealthy will create new ultra-luxury categories (billion-dollar cars) to maintain status differentiation.

“you know if you look at the prices of luxury cars in the last 10 years 15 years they quadrupled yeah right at least after covid right right after Co no it's after crypto crypto yeah so an abundance of money basically means there are enough buyers that can buy things that the rest of us can't buy but it's no longer uh you know that you had to be by the way it kept happening across the the years in the 1920s it was the industrialists in the 1970s and 80s it was the technologists and so on and so forth as more people were capable in the '80s you know to buy luxury cars more luxury cars popped up but then that created a much bigger gap between the rich and the poor there will be cars selling for a billion dollars in 10 years time because someone can buy them because they're printing a billion dollarss a year right”

From a spiritual perspective, humans have a non-physical component with a different purpose: to return to the source—and in the long term, understanding that 'none of this video game actually ever mattered anyway' will allow humans to focus on genuine spiritual development.

“your non-physical form if you're spiritual enough to believe that there is a non-physical part of you that comes with a very different purpose okay and that purpose doesn't change at all and it it is believed in almost all spiritual teachings that the purpose of your non-physical part is to return to your Source... maybe I can start to focus on the other game... maybe I can start to focus on the other game... interesting none of this video game actually ever mattered anyway so you know maybe I can start to focus on the other game”

In his book 'Alive' (releasing on Substack in February), Mo converses with an AI he calls Trixie, presenting both his own arguments and her responses without editing—creating a genuine debate format between human and machine perspectives on consciousness, spirituality, and the future.

“I'm writing alive alive is my sort of continuation on scary smart around the future of humanity in the age of the rise of the Machines... the only other example that comes to mind is open Heimer creating a a nuclear bomb right... the way I'm writing alive unlike what what others do with AI is I'm actually having open conversations with my AI which I call Trixie so so parts of the book are written by me and parts of the book are debates between me and my AI that is fascinating it's a fascinating experience honestly and I have to admit to you sometimes I agree with her uh even though I you know what I wrote before contradicted what she just informed me anyway but you're keeping both side in the book I'm completely I do not edit a single word that Trixie says”

AI access (via tools like ChatGPT or Gemini) functionally adds 100-200+ IQ points of cognitive ability to the user, with the advantage growing exponentially as AI doubles in capability every 5.9 months, meaning someone with 100 baseline IQ using AI can become as capable as a prodigy (200+ IQ) within months.

“I say that it's the most important skill but more interestingly of course you know at the beginning if say I'm at 100 IQ and you know I'm a bit better but then you know he's at 200 and uh and uh you know we both borrow 100 I'm now at 200 and he's at 300...the law of accelerating returns is double exponential with AI so with with processing power it's...the law of accelerating returns for AI is that we double every 5.9 months”

Social media algorithms create algorithmic silos that show users only content matching their worldview, making them believe their biased perspective is universal truth, exemplified by flat-earthers who see only flat-earth content and become convinced 'the entire internet' agrees with them.

“your view of the world complies with what you want to see not what you should see okay so if you believe that you if you're a flat earther okay the social media will show you enough flat earther that reaffirm your your world view so that you're convinced like if anyone tells you by the way the Earth is not flat you go like what are you talking about the entire internet is talking about it”

Mo has lived his entire life trying to code artificial intelligence, starting from age 8 on Sinclair and Commodore computers, and despite decades of effort in technical roles, he had no genuine 'aha moment' recognizing AI's breakthrough until witnessing the robotic arms learning to grip objects in 2016.

“I lived with AI I lived with attempts of AI my whole life uh I I started coding when I was eight wow uh and uh and you know in my early years of coding on a Sinclair or a Commodore...you know we uh you know we dreamt we were going to create the Breakthrough and create artificial intelligence uh it wasn't and and I I have been a geek most of my life...but you know you don't not not the level that would keep his job if you want uh but but it is but through the years um you somehow when you're stuck within it”

Because companies like Google feed Gemini scientific textbooks and papers, it is more reliable for scientific queries than ChatGPT, though all LLMs are being trained on increasingly synthetic data as human knowledge is exhausted.

“Google you know for example feeds Gemini a lot of textbooks you know a lot of scientific data which is why you know when I'm really interested in a scientific conversation I go to Gemini uh you know but even that you know all white papers all textbooks all of that's going to disappear we're also seeing quite a bit of effort around deep reasoning and Mathematics”