What this covers
Gopi Kallayil, Google's Chief Business Strategist for AI, frames artificial intelligence as a transformative technology comparable to electricity or the printing press—a cognitive amplifier rather than a threat to be feared. The conversation moves across the terrain of AI's economic potential, its reshaping of work and education, and the philosophical boundary between machine and human capability. Kallayil argues that the risks AI poses are neither unprecedented nor unmanageable; he draws an explicit parallel to how earlier transformative technologies—from fire to software—were weaponized yet contained through human-built systems. The centerpiece of his case is that an "AI-safety industry" will emerge organically, just as antivirus software did, to defend against malicious deployment.
The discussion spans concrete near-term applications and broader structural shifts. On employment, Kallayil characterizes job impact in three categories: some roles will be eliminated (as happened with buggy drivers and hand-loom workers), most will be transformed rather than erased (bank tellers becoming relationship managers after ATMs), and new categories will form that do not yet exist. He identifies the most promising frontier not as AI built by technologists but as entrepreneurs in non-tech fields—education, health, food services—using AI as a tool to solve domain-specific problems. On learning, he envisions a move from lockstep mass education to personalized instruction that adapts to each student's pace. Throughout, he emphasizes that consciousness, emotion, and love remain uniquely human properties that no AI system will embody, however fluent it becomes at language. The macro-economic forecast he cites—roughly $15 trillion in potential GDP impact across generative and analytical AI—rests on the premise that widespread adoption will drive productivity and prosperity gains at a scale comparable to what lifted a billion people from poverty in the last three decades.
Gopy Kle argues that AI represents a foundational cognitive amplification technology comparable to the printing press and electricity, with transformative potential for solving human problems through entrepreneurial applications, but requiring proactive governance safeguards and broad education to ensure beneficial outcomes while mitigating misuse risks.
- AI amplifies cognitive capability rather than physical capability, marking a qualitative shift from previous technology waves
- Historical precedent (fire, aviation) shows humans will both use and misuse any powerful technology, requiring governance structures like anti-virus software for malicious AI
- Economic projections (McKinsey: $3.3T from generative AI, $11.8T from analytical AI) suggest massive productivity gains creating new job categories while disrupting existing ones
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
AI differs from previous transformational technologies because it amplifies cognitive capability rather than physical capability, making it fundamentally different from cars, construction equipment, or the industrial revolution which primarily enhanced physical capabilities.
“But what is different each this time is the earlier waves of technology primarily helped us with physical capabilities like cars moved us and goods or construction equipment helped us build tall buildings industrial revolution etc. But AI is all about cognitive capability.”
YouTube has democratized content creation by allowing any individual to become an independent creator and publisher of content, especially video content, removing barriers that previously required relationships with broadcasters and managers.
“I've always been fascinated by YouTube as a platform and I'll tell you why. It allowed for human beings all anybody who chose to to become an independent creator publisher of content especially video content. It's really democratic in that sense. Correct. Two billion two billion people watch it. In an earlier era, if you wanted to talk show on video, you needed to talk to the right broadcasters etc. Now you can use YouTube if you're talented or a musician. You don't need a manager and a gig, an agent.”
Humanity still has many unsolved problems in the human condition—housing, hunger, quality healthcare, education, and social justice—and the question of how humans will use AI entrepreneurially to solve these problems is what excites Gopy most.
“For me, the thing that I'm most hopeful about Marian and Andrew is I feel that even after all these years of progress, there are still so many problems in the human condition that we still have not successfully solved for as a specy, right? Whether it's housing or hunger or good quality health care for everyone and and education, social justice, etc. How will humans use in an entrepreneurial way the potential of AI to solve for this is what excites me the most”
The danger that powerful technology will be used for destructive purposes is not unique to AI; throughout history every beneficial technology (fire, planes, software) has been weaponized, and humans have responded by building checks and balances (fire detection, firehouses, arson laws, antivirus industry), so AI will follow the same pattern.
“Every single piece of technology that humans have used to greater good, we also use it for devious destructive purposes... So over the years we built checks and balances to protect ourselves from it.”
Unlike earlier technology waves that amplified physical capabilities (cars, construction equipment, the industrial revolution), AI is fundamentally different because it amplifies human cognitive capability.
“the earlier waves of technology primarily helped us with physical capabilities... But AI is all about cognitive capability.”
The automated teller machine (ATM) transformed the bank teller's job from cash counting and distribution to customer relationship management rather than eliminating the role, providing historical precedent for how technology changes rather than necessarily eliminates jobs.
“Just like thanks to the automated teller machine, the bank teller's job changed and became a customer relationship manager as opposed to counting cash and handing it to you when you needed to withdraw.”
Every single powerful technology that humans have created, including fire, planes, and software, has been used by humans for both beneficial and destructive purposes, requiring society to develop corresponding safeguards, checks, and balances.
“Every single piece of technology that humans have used to greater good, we also use it for devious destructive purposes. That is an incredible human talent.”
The 2.6 million applications available on the Android Play Store and Apple App Store demonstrate the value of platform ecosystems, where the platform's usefulness derives not from the core capability but from the proliferation of specialized applications users can choose from based on their individual needs.
“What makes it useful is not the fact that even though we call it a mobile phone, it's not the fact that you can actually call someone with it. That's useful. But the fact that I can download from a menu of 2.6 million applications available on the Android Play Store or the equivalent in the Apple Play Store. And if I look at your phone Andrew and Mariana's phone, it'll look different because you have chosen which of those applications are most useful for your life depending on are you interested in music or exercise or cooking or travel or or learning a language”
YouTube creators produce extensive educational content on problems (like appliance repair error codes) in multiple languages and at no cost, demonstrating human willingness to help others without payment, which Gopy calls 'gifting'.
“There is no problem that I run into these days that I can't find an answer on YouTube. And I'm going around the house and something's not working. And I'm like, this washing machine, I'm not familiar. What does error code 402 mean? I go to YouTube. There's somebody who has taken the time to record a video explaining error code 402 and how to fix it.”
Jobs impacted by AI can be categorized into three groups: drained jobs that disappear entirely, changed jobs where the work continues but how it's done transforms, and gained jobs that are newly created, with the vast majority of jobs falling into the 'changed' category.
“The way to look at job impact from AI, I would say, is you have to put them into three categories. There will be some jobs that will be what I call drained. They'll go away. And that is the case with any technology platform shift. When we went from horsedriven drawn buggies to cars, it happened. When we moved from hand loom to textiles through machinery, that happened. And when we moved from very very primitive forms of agriculture to agriculture through automation tractor and all harvesters that happened as well. So every wave of technology shift that always happens uh or printing when we went from the manual composing and printing of material to digital publishing. The type setters job went away. So there will be just set of jobs that will be drained. There will be a vast majority of jobs that will get changed. The jobs will continue. But how we do it, what we do with it will change it.”
While AI systems can beat the best players at chess or Go through self-play and learning, this does not constitute true intelligence because humans can explain their moves logically and create games from scratch, capabilities AI systems have not yet demonstrated.
“it is true that we build AI systems that can beat the best player of chess or a go and at that point you feel like they are smarter. That's because there is components of the eye software that allows for it to play against itself, learn and and build and and do things from a game strategy. But I don't quite buy into theory that it is smarter because what distinguishes humans from machines in that case or an AI system is the human can still explain logically certain moves or even how create the game from scratch”
AI skills can be understood through the electricity analogy with three tiers: people who simply use AI safely (like flicking a switch), people who build things using AI in their own domain (like hospitality or airlines), and people who build the technology itself (requiring deep computer science, neural networks, and deep learning knowledge).
“like I would say with electricity some people only use it they just know how to flick a switch... There are others who work on electricity generation... and there are others who build equipment that runs off electricity... In the same way with AI”
The most exciting frontier of AI is not building AI apps themselves but entrepreneurs in non-technology fields (education, health, food services) using AI as a component to solve problems in their core business, because AI amplifies cognitive capability across every field of human endeavor.
“the core business is not about technology or AI but it can be completely different area but uses AI to solve for a problem and that's where I think there'll be some really incredible examples of how AI is deployed”
AI's impact on jobs falls into three categories: jobs that will be drained (eliminated, as happened with buggy drivers, hand-loom workers, and typesetters), jobs that will be changed (the vast majority—e.g., bank tellers became relationship managers after ATMs), and jobs that will be gained (entirely new categories that don't exist today, like web analysts and data scientists created by the internet).
“There will be some jobs that will be what I call drained. They'll go away... There will be a vast majority of jobs that will get changed... And then there'll be an entire category of jobs that get created that are jobs that will be gained.”
YouTube democratized content creation by letting anyone become an independent publisher without gatekeepers, and a notable category of creators voluntarily and for free solve problems for strangers (repair guides, recipes, lessons); Kile expects a similar explosion of entrepreneurial creativity using AI as a tool to solve problems across every field.
“It allowed for human beings... to become an independent creator publisher of content... A similar kind of explosion is what I'm expecting to see. Entrepreneurs using that human creativity and imagination will look at AI as a tool”
AI enables a shift from mass education to custom education where instruction adapts to each learner's ability and pace, freeing students from the lockstep model in which a whole class moves at the same speed toward a common exam.
“Same way in education. Today we do mass education but we could actually do in the future with AI custom education where the education is adapted to the learning ability of each person and at a pace at which you want.”
AI was first termed in 1960 by Dartmouth mathematician John McCarthy, lay dormant as a field because it was dismissed as theoretical with no practical value, became marginalized in academia where pursuing AI research could harm careers.
“the term was coined as think I think in 1960 by uh the Dartmouth mathematician John Murray. So we were on in fact I studied AI in graduate school but back then it was dismissed as a interesting theoretical concept but of no practical value because we can't implement it. So in fact if you did a PhD in certain fields of AI like neural networks etc. you would be considered that your career is not going to be anywhere. You could not get a research projects or get hired into university.”
Google has demonstrated an AI application called 'shopping try on' that allows users to photograph clothing on another person or in images, photograph themselves, and see how they would look wearing that outfit through AI image synthesis.
“We through four weeks ago, we showed a really cool demo where I can actually point to what Andrew is wearing, take a picture of it on my phone, take my picture, and combine the two using an AI. All of a sudden, I'll be wearing Andrew's shirt. Or it would look like how I would look in the outfit Andrew is wearing.”
The single most important AI skill to cultivate—especially in children—is curiosity: educate yourself about available tools, play with the free ones daily, and treat learning as a game (e.g., learn one new AI capability each day and teach the family).
“the most important thing I'll tell them is be curious, educate yourself about what is available now. Start playing around with the tools because many of them are free”
Applications are the most exciting next stage of AI development, with hundreds of startups and established firms racing to create better, cheaper, and more sophisticated AI applications.
“You know, the conversation about AI is increasingly centered on the importance of applications that you can derive. Hundreds of startups and established firms are racing to create better and cheaper, more sophisticated apps.”
If you did a PhD in fields like neural networks in the 1980s-2000s, you would be considered unhirable and unable to get research funding or university positions—the field was academically radioactive until the last 10 years.
“So in fact if you did a PhD in certain fields of AI like neural networks etc. you would be considered that your career is not going to be anywhere. You could not get a research projects or get hired into university.”
Self-driving car adoption will displace the jobs of people whose primary function is driving cars for a living, though this is not unique to AI and has occurred with every technology platform shift in history.
“So there the question always is is it taking away the job of of some people who drive cars for a living whose job is to drive cars and that is indeed true. If the whole world switch to self-driving car that job will go away.”
AI's interface has evolved rapidly from search engines to answer engines to conversation engines, and within months has moved further to 'task engines' that autonomously perform tasks and 'intelligence engines' that provide context, while increasingly acting as a remote control to the physical world by perceiving and interpreting objects and completing real-world tasks.
“we actually moved past that. We moved past to now what I call task engine where they're autonomously doing tasks for you and they've actually moved to intelligence engine... they're becoming a remote control to the physical world”
In the last 30 years, 1 billion people were lifted above the poverty line, primarily from economic development in India, Southeast Asia, China, and Brazil, demonstrating the human benefit of economic growth.
“And I would endorse that in the last 30 years, we lifted 1 billion people out above the poverty line as defined by the United Nations. And it came mostly from growth in economic development in India, Southeast Asia, China, Brazil.”
Love, emotion, and consciousness are uniquely human qualities that AI systems will never be able to replicate, even though AI can simulate romantic language by recombining text learned from novels and movie scripts.
“there's another aspect of the human condition that I personally believe that we will never ever see the smartest AI system ever be able to bring up and that is the concept of love and emotion and consciousness that is a unique human quality even though you hear stories of AI systems acting like that they are in love with the person that they're talking to. But I think in that case that's because they're making up words and sentences that they must have read and learned in romantic novels and movie scripts etc. So the AI systems know how to respond to particular question or emotion but they're not feeling the emotion the way to there is no consciousness there.”
Every hundred years or so, humans develop a transformational technology that completely changes the arc of civilization, with examples being the printing press around 1640, electricity around 1750, and now artificial intelligence.
“Every hundred years or so, humans will come up with a piece of technology that completely changes the arc of civilization. For example, around 1640 or so is when Yseph Gutenberg invented the printing press. And 600 years later, books and other printed materials still are such a significant part of information creation and dissemination.”
AI systems like Gemini Live can now see objects in a user's physical environment, interpret them, have conversations about them, and complete tasks by providing specific instructions (like 'use the cumin seed' or 'put that garam masala into the pan') based on visual understanding.
“on this phone I could do a video audio demo and your audience may be able to hear it actually can see the objects around me and interpret it for me. I have this bowl of food and if I hold my Gemini live with a video view, I can have a conversation with it. It'll explain to me what nutrition is in the food and then uh in certain cases it can then go on to complete tasks for me etc. So it is astonishing that it's moved from just a search engine where to discover some things and then I have to interpret it. Now it can help me complete a whole task like repair a bicycle or cook a dish by interacting with the world and say use the cumin seed or put that garam masala into the pan.”
Self-driving cars (specifically Waymo, a sister company of Google) have completed over 10 million rides, equivalent to the entire population of Michigan or North Carolina having used the service, and are commercially available in San Francisco and Los Angeles with limited service in Phoenix and coming soon to Atlanta, Miami, and Tokyo.
“it's already available in San Francisco and Los Angeles for anyone who chooses to ride in them and limited bases in I think Phoenix and coming soon to Atlanta, Miami and Tokyo for international expansion.”
What surprised Gopy most about AI is the speed with which it emerged from academic dormancy and became mainstream—something that has occurred in rural India where extended family members now ask him about AI despite having had no prior exposure to technology.
“The number one thing that has surprised me is the speed with which it suddenly came out of the box and has become such an incredible part of our awareness. So AI is not a new field and the term was coined as think I think in 1960 by uh the Dartmouth mathematician John Murray.”
NotebookLM is an AI-powered product that allows users to upload documents and materials about a topic, creating a personalized knowledge repository that can answer questions based only on the uploaded documents and generate audio podcasts about the material.
“notebook LM which is an AI powered product. It's your project notebook, your project site. You upload a lot of documents and material you have about AI. And actually, it's likely somebody has created a notebook about AI that's publicly accessible. And what it does is it's almost imagined like you have a personal friend except in this case an AI powered tool that knows everything about a particular area of AI that you wanted to know about. Instead of just whatever is available on the web, you can limit it, contain it, saying, I only want you to look at these eight textbooks about AI and then answer questions I have as I learned. Teach me about it and you can ask questions. It'll create a audio podcast if you want and you can listen to it.”
Travel is consistently the first industry to adopt a new technology platform shift (as seen with websites and mobile phones, and now AI), because the purchase decision and intermediate stages are digital and non-physical even though the product is experiential; retail and e-commerce follow closely behind.
“whenever there's a technology platform shift Mariana travel is the one that first you'll see the adoption with websites you saw that and with phones also”
Just as the antivirus software industry emerged to defend against malicious software, an entire industry category of defensive AI—AI software that protects against malicious AI—will emerge to put boundaries around the technology.
“AI software that will defend you protect you against malicious AI software will be an industry category that will emerge that'll put some boundaries around this”
The belief that AI will eliminate the need for human coders is false, because AI will assist with and do some coding but humans will still be needed as an adjunct (some code can only be written by humans at this stage), and the world's demand for code will be so large that even machines plus humans will fall short—analogous to perpetual energy shortages.
“this thing that we'll have AI and we will no need anyone to code I don't think is going to be true... the amount of AI and other automation software that'll go into every devices... there will be so much code that will be required”
Just as every city has fire detection systems and 24/7 firefighting infrastructure, and there are international bodies like the International Atomic Energy Commission, there will need to be an international AI protection commission and AI watchdog bodies to manage AI risks—a regulatory framework that does not yet exist.
“Really up to us to put all of that in place. So for example, just like we have the firehouse in every city, we don't yet quite have the AI watchdog bodies or like we have the international atomic energy commission. We don't yet have the international AI protection commission. I'm also predicting that and this actually is not my original thought.”
Love, emotion, and consciousness are uniquely human qualities that will never be embodied in an AI system; apparent AI expressions of love are merely recombinations of words learned from romantic novels and scripts, with no genuine feeling or consciousness behind them.
“the concept of love and emotion and consciousness that is a unique human quality even though you hear stories of AI systems acting like that they are in love... they're making up words and sentences that they must have read and learned in romantic novels... there is no consciousness there”
AI may seem smarter than humans in narrow domains (beating the best chess or Go player) but is not genuinely smarter, because humans can still logically explain moves and create games from scratch, and no AI system has yet demonstrated the ability to invent a wholly new mathematical theorem the way a human researcher can.
“I don't quite buy into theory that it is smarter because what distinguishes humans from machines... the human can still explain logically certain moves... no AI system is yet shown the capability to come up with a completely new theorem of math”
Generative AI is estimated to add about $3.3 trillion in economic impact (slightly larger than the UK economy) per a McKinsey bottom-up analysis of 57 industries, while analytical AI/machine learning is projected to generate roughly $11.8 trillion—four times that—so AI could add about five UK economies of GDP growth, which Kile argues is good for jobs and global prosperity.
“the impact of generative AI over the next few years is estimated to be about 3.3 trillion... that's slightly bigger than the UK economy... AI systems on that side are supposed to generate 11.8 trillion in GDP growth, four times that number”
The vast majority of the global population is not adequately educated about AI, including government bodies, regulators, and policymakers who must make decisions about AI governance despite the technology moving faster than education can keep pace.
“No, I don't believe they're educated enough. a few group of people close to the technology you know AI elite who understand it but the vast majority of people who have to deal with regulations and you know government bodies who have to take decisions and policy makers etc don't understand it and they need to go through an education process because the technology is moving so quickly”
In the last 10 years people became intrigued by AI possibilities, in the last 24 months AI suddenly exploded, and in the last 6 months it became a hot topic in elections with even people on the street having heard of it.
“So it lay sort of dormant till in the last 10 years people in the field were getting intrigued by it and the possibilities but in the last 24 months it suddenly just exploded and in the last 6 months it has become the hot topic that elections have fought on and and everyone is asking the question and people on the street have at least heard of some perspective on it.”
Electricity provides an analogous framework for AI skills: some people only use it and know how to flip a switch safely, others work on generation/distribution/transmission, and others specialize as electrical engineers; similarly, AI skills will range from users to builders to specialists.
“I would say with electricity some people only use it they just know how to flick a switch and turn on a device and they're full beneficiaries of it but they know how to handle it safely. Beyond that they don't need to do it. There are others who work on electricity generation, distribution, transmission and there are others who build equipment that runs off electricity and specialize in that electrical engineers of sorts. In the same way with AI uh it depends on where they want to go.”
The most important skill for children to develop in an AI world is curiosity and the ability to educate themselves about available AI tools, with hands-on experimentation and learning by doing.
“Having said all that, a fundamental rudimentary understanding of what is AI, how to use it, where you can apply it, how it is changing. Being curious about it is essential. So the most important thing I'll tell them is be curious, educate yourself about what is available now. Start playing around with the tools because many of them are free and available.”
Henry Kissinger said in an interview that AI is what keeps him up at night, and he stated that in the future, countries will not fight over territory but over data.
“I was lucky enough on Henry Kissinger's 100th birthday, we had a party for him at CSIS, and I was able to interview him and I asked him what kept him up at night. And without hesitation, he said AI. And then he said something that was kind of chilling to me. He said, 'In the future, countries are not going to fight over territory. They're going to fight over data.'”
An entire anti-virus software industry emerged to defend against malicious software, creating boundaries that allow software to be used safely; a similar industry category will emerge for defending against malicious AI systems.
“So to defend against that an entire class of industry called the anti virus software industry has emerged which we installed this software it protects our phones protect our computers similarly AI software that will defend you protect you against malicious AI software will be an industry category that will emerge that'll put some boundaries around this so we can use these technologies in a more safe uh and protected way”
Productivity improvements from AI tools lead to economic development and upliftment, with McKinsey estimating generative AI will add approximately $3.3 trillion in economic value, roughly equivalent to the entire UK economy.
“Whenever we build these tools, Mariana, it improves productivity right things get done faster differently and anytime there is productivity improvements, it leads to economic development upliftment and I have great expectations from AI. I'll give you some numbers that I've seen. uh I used to work at Mckenzie before and Mckenzie put out a report called the economic impact of generative AI and in that they say the impact of generative AI over the next few years is estimated to be about 3.3 trillion they looked at 57 industries and did a bottoms up calculation that's a large number let me give you a sense of scale that's slightly bigger than the UK economy”
A similar explosion of entrepreneurial creativity to what YouTube enabled will occur with AI, where entrepreneurs use AI as a tool and ask 'what problem can I solve?' leading to proliferation of successful entrepreneurial ideas across music, education, health, food and other fields of human endeavor.
“A similar kind of explosion is what I'm expecting to see. Entrepreneurs using that human creativity and imagination will look at AI as a tool and say what problem can I solve for? And you will see a proliferation of ideas, entrepreneur, successful entrepreneurial ideas in every field of human endeavor from music to teaching music to uh understanding music or in health or in food.”
Humans created AI through natural intelligence and therefore humans have agency and control over it—we can decide how to use it and put checks and balances in place because we built it.
“at the end of the day, who created that AI? It is humans. Our natural intelligence created it. So we have some agency over it, some sense of control over it and we can decide how we want to use it and put the checks and balances etc. It is up to us to build it.”
Humans have a sense of connection with spirituality or religion that constitutes a unique awareness connection to the higher self, which Gopy expects no AI system will ever be capable of replicating.
“not to mention humans have a sense of a connection with the higher self and we call it spirituality or or might be called religion for lack of better word. But whatever their definition, each person has their own individual definition. That awareness connection, I never expect any AI system to be capable of doing it. That will be uniquely ours.”
AI and other coding automation will not eliminate the need for human coders because the demand for software in devices (watches, cars, dashboards, phones, washing machines, doorbells, thermostats) will be so vast that AI plus human coding combined will be insufficient to meet demand, similar to energy shortages.
“I also think that the need for code in the world will be so much that we'll need machines plus humans to fulfill it and even then there'll be a shortfall much like energy right we don't just seem to be able to generate enough energy for the world's needs we constantly in a hunger mode and same way the amount of AI and other automation software that'll go into every devices that we seeing it's in a watch in our car in our dashboard in our phone in our washing machine our doorbell in our thermostat so there will be so much code that will be required said that I think AI plus humans will always be there in the coding business for a long time.”
The human quality of willingness to help others for free without transactional reward may distinguish humans from machines, even if machines become smarter in specific technical domains.
“Is that what will distinguish humans from machines? I mean because machines are going to be much smarter than us but are will they be willing to cooperate and to help other machines or other humans as you know as this that we are describing.”
The medical field provides a precedent where experts explain complex technologies and treatments to decision-makers and the public—a model that will be necessary for AI governance, though it will slow down decision-making.
“there are things that is happening in the medical field that very few people actually understand what we goes on in terms of heart surgery or things we do in the brain or around uh different kinds of pharmaceutical development or genetic modification so many aspects of it and we rely on a few experts to explain it to the people who have to make a decision it is slows down the process but that's how we solve for the problem and the same thing will apply to AI as well”
Travel is the first industry to adopt major technology platform shifts like the internet and mobile phones, and is now the first to adopt AI, because the buying process is primarily digital with minimal physical involvement before purchase.
“The first one I always see whenever there's a technology platform shift Mariana travel is the one that first you'll see the adoption with websites you saw that uh and with phones also with mobile phones and the same thing I'm beginning to see with an AI. It's just the nature of the product. The product itself might be physical and experiential, but before you actually buy that plane ticket or the hotel stay or the vacation package, the intermediate stages for a consumer are very easy. It's digital.”
In advertising, AI allows creation of 3 million versions of a campaign fine-tuned for every micro market, compared to the previous digital era where one campaign version reached the entire world.
“in the past I'd have used digital to do one digital campaign that the whole world saw now I can do 3 million versions of it because AI helps me fine-tune it for every micro market.”
AI-powered medical care could provide personalized treatment by researching all available data about individual patients to create a comprehensive picture, enabling different levels of care than currently possible.
“This will be doctors who will know how to use AI without needing to know the underlying technology because they left other people do it. But I'm going to use AI to give a different level of care for Andrew and for Mariana that I could not do before where I can actually research everything about them and the data points that I have access to in order to put together a full composite picture of the patient in front of me.”
AI could enable custom education where students progress at their own pace rather than forcing uniform classroom pacing, with some students taking 8 months for calculus while others take 8 days, supported by AI tutors.
“Today we do mass education but we could actually do in the future with AI custom education where the education is adapted to the learning ability of each person and at a pace at which you want. Today we insist that Mariana if you and Andrew are in the same classroom in the same uh program uh whether you're doing Spanish or business engineering you have to pace with each other. The whole class moves at the same pace and the end of the term you get an exam. What if thanks to AI if I say I'm going at a different pace I'll have the freedom to say I'm going to take eight months to learn calculus course that a brilliant mind like Andrew can do in 8 days and that's totally fine and he can take his exam after 80 days and I will do it after 8 months because that's a pace at which I want to go and I have an my regular teacher plus an AI enabled special tutor for me that allows me to pace myself at my pace.”
The best way to learn about AI is non-linear and AI-itself-mediated: use tools like Google's 'Learn About', converse with AI tutors (OpenAI, Perplexity, Gemini), use deep research options, and use NotebookLM to constrain answers to chosen source documents and generate audio podcasts, while still relying on a human 'tribe' of curators for the best-quality content.
“Use AI to explore it in a nonlinear way in various paths you want to take... I still use humans... they are sort of like my human editors and curators for the best content”
Finding a group of close friends or followers who are knowledgeable about AI and regularly discuss and share content about it helps curate the best quality content through human editors rather than relying solely on AI or impersonal algorithms.
“I still use humans I have a group of close friends who are who are avid readers and discusses and we have a small discussion group. I track what they post. They'll find links and I'm like I have no idea where they get these sources from. So they are sort of like my human editors and curators for the best content. So you should find a good tribe. Find four friends uh or followers. If you don't know them, find them on the web and and they are regularly talking about it, posting about it on and for follow them on social media and they become your human editors for the best quality content you need to know about AI.”
Learning about a topic using AI tools in a non-linear way, exploring various paths and following curiosity down unexpected tunnels, is more fascinating and effective than linear reading or discrete bite-sized information.
“Use AI to explore it in a nonlinear way in various paths you want to take and stumble down various paths that you did not even know existed and you discover tunnels and maces down the AI learning path and the AI tool leads you and let your curiosity guide you that'll be much more fascinating interesting way to learn”
Technology is transitioning from search engines (requiring user interpretation) to answer engines (providing direct answers) to conversation engines (enabling dialogue) to task engines (autonomously completing tasks) to intelligence engines (providing contextual intelligence and acting as a remote control for the physical world).
“I did talk about search engine to answer engine to conversation engine because that's what AI helped us do. We had gone from search to conversation but we actually moved past that. We moved past to now what I call task engine where they're autonomously doing tasks for you and they've actually moved to intelligence engine where we simply moved from just being a tool to get information to actual intelligence. Right? it puts us context and the other thing they're beginning to do is actually they're becoming a remote control to the physical world.”
Before releasing products, Google has teams that check to ensure products follow established guidelines and principles—a practice that slows product development but provides safety benefits and is the right decision if safety is the priority.
“So within the industry for example before you release a product there are groups teams that will actually check to make sure that the products you're releasing follow certain guidelines and principles that you establish. It does slow things down for product development perspective but if that is the safer thing to do than make a mistake and release it and if that's the right decision then you you you should do it to build some levels of safety.”
Many scenarios being portrayed about AI behaving dangerously in the future are not actually happening, which means the trajectory of AI could go in many different directions and is not predetermined.
“my personal thinking around this and a lot of it I would say speculation because we're seeing if AI could behave like this in the future is many of the scenarios that are being portrayed are not happening. So, this could go in many directions.”
AI is not a new field (the term was coined around 1960 at Dartmouth) and was long dismissed as a theoretical concept with no practical value—pursuing a PhD in areas like neural networks was once considered career-ending—yet it lay dormant until interest grew over the last decade, exploded in the last 24 months, and became a mainstream topic in the last 6 months.
“the term was coined as think I think in 1960 by uh the Dartmouth mathematician John Murray... if you did a PhD in certain fields of AI like neural networks etc. you would be considered that your career is not going to be anywhere”
Over the last 30 years roughly 1 billion people were lifted above the UN-defined poverty line, driven mostly by economic development in India, Southeast Asia, China, and Brazil; if AI drives similar productivity and growth, it can produce comparable prosperity gains for the world.
“in the last 30 years, we lifted 1 billion people out above the poverty line as defined by the United Nations. And it came mostly from growth in economic development in India, Southeast Asia, China, Brazil.”
AI represents the birth of one of the most profound technologies ever created, comparable to the printing press (c. 1640) and electricity (c. 1750), each of which appears roughly every hundred years and changes the arc of civilization.
“Every hundred years or so, humans will come up with a piece of technology that completely changes the arc of civilization.”
Most people—including policymakers, regulators, and government bodies who must make decisions—are not educated enough about AI, and even industry insiders feel they fall behind every few weeks because innovation is moving so fast; this knowledge gap will slow the policy process much as complex medical fields rely on a few experts to inform decision-makers.
“No, I don't believe they're educated enough... the vast majority of people who have to deal with regulations... don't understand it and they need to go through an education process because the technology is moving so quickly”
Ethan Mollick from Wharton has written a book called 'Co-Intelligence' and regularly posts about AI on LinkedIn, Medium, and Substack, providing accessible regular content about AI for those wanting to stay current.
“both in terms of a book and a regular uh post comes few times a week. Uh I went to Wharton and Wharton professor Ethan Mollik has been doing a great job. He wrote a book called Co- Intelligence. He's also got LinkedIn post and medium post uh sorry Substack that you can follow.”
Electricity around 1750 is a foundational technology that has entered every aspect of human life, visible in at least 15 use cases in any room, and its interruption for 10 seconds would cause civilization to grind to a halt globally.
“same way 1750 to electricity is so foundational technology that it has entered every aspect of our life. If you just stop and look around the room you're in, you'll find at least 15 use cases. Uh we our cars drive on it. We come up on elevators. Every equipment in the kitchen seems to run on it. It's it's and if you turn it off for 10 seconds around the world, civilization will grind to a halt.”
Aircraft have been weaponized to transport weapons of mass destruction despite being primarily useful for transporting people, goods, medical equipment, and medicines—another example of technology dual-use.
“The same thing with planes. Planes useful to transport people and goods uh and medical equipment and medicines and yet we some humans said this is also a good way to transport weapons of mass destruction.”
Because humans created AI using natural intelligence, humans retain agency and control over it and can decide how to use it and what checks and balances to impose.
“who created that AI? It is humans. Our natural intelligence created it. So we have some agency over it, some sense of control over it”
Waymo self-driving cars (a Google sister company) have surpassed 10 million rides—equivalent to everyone in Michigan or North Carolina having ridden one—and are commercially available in San Francisco and Los Angeles with expansion to Phoenix, Atlanta, Miami, and Tokyo, demonstrating an incredible pace of adoption.
“We've crossed 10 million rides now. 10 million rides is the equivalent of everybody in the state of Michigan or North Carolina having ridden one.”
AI-powered virtual try-on (e.g., Google's 'Shopping Try-On') already exists, allowing a consumer to photograph an item and combine it with their own photo to see how a garment would look on them.
“It's called shopping try on... four weeks ago, we showed a really cool demo where I can actually point to what Andrew is wearing, take a picture of it on my phone, take my picture, and combine the two using an AI.”
A tool called 'learn about' from Google builds custom lessons when given a topic (e.g., 'artificial intelligence', 'generative AI'), allowing structured AI-guided learning.
“There is a tool, it's an experimental tool. If you search, it's from Google. It's called learn about. So you go to the tool called learn about. It's on on the web. And then you can say I want to learn about artificial intelligence or I want to learn about generative AI and it starts building a lesson for you and doing that.”
The book 'Genius Makers' by Kate Mets is the best book for understanding how AI was developed and who were the key people involved in its development.
“I was thinking about it. I didn't want to, but among all of the books that I looked at, Genius Makers is the book that I liked the best by Kate Mets. um it really tracks how AI got developed, who are the people and uh I find it really fascinating read. So I would recommend Genius Makers.”
Humans have been building tools throughout the 12,000 years of recorded history to solve problems and improve quality of life, starting with physical tools like water bottles and now transitioning to a cognitive tool (AI) that amplifies mental capability rather than physical capability.
“If you look at the last 12,000 years of human history, humans have built tools one after the other. We keep building tools. This bottle of water in front of me is a tool. So I can carry water warm or cold or with me where I want to and deal with my thirst. So we are very good at building these tools to solve problems that improve quality of life for everyone and mostly we did with physical tools but now we're dealing with seeing the birth of a tool that will amplify our cognitive capability.”
Gopy is working on a book about AI and expects to publish it in approximately three months, though details cannot be shared until publication is finalized.
“I forgot to tell you this morning, I spoke to a publisher. So, if everything goes through in three months from now, if you ask me the answer, I'll say read my book about it. But right now, I can't.”
Gopy's own music and talks are on YouTube with significant views, and he expresses gratitude that people in places like Latvia or Germany might benefit from his ideas without his knowledge, suggesting that he finds personal fulfillment in helping unknowns.
“Even my music is out there and it's got a huge amount of views. A lot of my talks are there and people are watching and commenting on it. I have no idea they are, but I'm grateful if they can benefit from my ideas, thoughts on music sitting somewhere in Latvia or or Germany. It makes me happy.”
Electricity is so foundational that if turned off worldwide for 10 seconds, civilization would grind to a halt, and any room contains at least 15 use cases of it.
“if you turn it off for 10 seconds around the world, civilization will grind to a halt”