[0:00:00] More of the world's intellectual capacity could reside inside of data centers than outside of them. This is an extraordinary statement to make, and of course we could be wrong, but I think it really bears serious consideration. A superintelligence, at some point on its development curve, would be capable of doing a better job being the CEO of a major company than any executive, certainly me, or doing better research than our best scientists. As we prepare for this possibility, we are guided by three core beliefs.
[0:00:34] Number one, we believe that democratization of AI is the only fair and safe path forward. Democratization of AI is the best way to ensure that humanity flourishes. On the other hand, centralization of this technology in one company or country could lead to ruin. The desirable future a couple of decades from now has got to look like a world of liberty, democracy, widespread flourishing, and an increase in human agency.
[0:01:04] Some people want effective totalitarianism in exchange for a cure for cancer. I don't think we should accept that trade-off, nor do I think we need to. AI should extend individual human will. We'll probably need superintelligence to help us figure out the new governance mechanisms to ensure that this happens fairly at scale and to avoid problems like extremely unbalanced compute access or something else. Second, we believe that AI resilience is a core safety strategy.
[0:01:35] We don't mean that this is the only safety strategy. We will continue to need to build safe systems and solve difficult technical alignment challenges. But increasingly, we need to start broadening how we think about safety to include societal resilience. No AI lab, no AI system can deliver a good future on their own. For an obvious example, there'll be extremely capable biomodels available open source that could help people create new pathogens.
[0:02:02] We need a society-wide approach about how we're going to defend against this. And third, the future of AI is not going to unfold exactly like anyone predicts. And we believe that many people need to have a stake in shaping the outcome. The development of AI has already held many surprises, and I assume there are bigger ones to come. We understand that with technology this powerful, people want answers. But it's important to be humble about what we don't know and always remember that sometimes our best guesses are wrong.
[0:02:34] Most of the important discoveries happen when technology and society meet, sometimes have some friction, and co-evolve. For example, we don't yet know how to think about some superintelligence being aligned with dictators in totalitarian countries. We don't know how to think about countries using AI to fight new kinds of war with each other. We don't know how to think about when and whether countries are going to have to think about new forms of social contracts. But we think it's important to have more understanding in society-wide debate before we're all surprised.
[0:03:07] Of special note, and related to all three points, we continue to believe that iterative deployment is a key strategic insight, and that society needs to contend with and use each successive new level of AI capability, have time to integrate it, understand it, and decide how to move forward. This has been working surprisingly well so far. If we are right, and systems continue to improve at this pace, it's going to change the economics of a lot of things.
[0:03:36] A really great thing about AI progress is that it looks like many things are going to get much cheaper and have much faster economic growth. We're already seeing what AI is doing for access to high-quality healthcare, education, and more. In the coming years, we expect to see robots make many products and physical goods cheaper as supply chains get automated. The limit to how far this cost reduction can go may only be government policy. But the other side of this coin is that current jobs are going to get disrupted as AI can do more and more of the things
[0:04:08] that drive our economy today. It'll be very hard to outwork a GPU in many ways. It'll be easy in some other ways. For example, we really seem hardwired to care about other people much more than we care about machines. We're somewhat less concerned about the long-term future. Technology always disrupts jobs. We always find new and better things to do. The people of 500 years ago would have thought that our current jobs often look silly, like ways to entertain ourselves, create stress. And the people 500 years from now
[0:04:39] Hopefully we'll look at us, hopefully look to us, like impossibly rich people playing games, trying to find ways to pass their times. But we should all hope they feel much more fulfilled than we do today. I'm confident we will keep being driven to be useful to each other, to express our creativity, to gain status, to compete, and much more. But the specifics of what we do day to day will probably look very different. Each generation has built on the work of the generations before.
[0:05:08] And with new tools, the scaffolding gets a little taller. This collective external lattice, the set of tools that we have built up around ourselves is remarkable. And we are capable of doing things that our great-great-grandparents couldn't have dreamed possible. It is a moral imperative to make sure that our great-great-grandchildren can stay the same. And technology, and especially AI, is how we're going to get there. For a democratic AI future, it is not enough to just give people tools and wealth. We also need to give them agency and power.
[0:05:42] The vision that AI companies lay out fundamentally reduced to either unilateral control or decentralized power. Sharing control means accepting that some things are going to go wrong in exchange for not having one thing go mega wrong. Cemented totalitarian control. This is a fundamental trade-off of democracy, and it is one that we believe in very strongly as the way to give everyone collective agency over the future. Of course, this is not to suggest that we won't need any regulation or safeguards. We obviously do urgently.
[0:06:13] Thank you very much. It's really a treat to be here in India.
[0:06:47] And it's incredible to see the country's leadership in advanced AI. I was last here a little over a year ago, and it's striking how much progress has happened since then. We've gone from AI systems that struggled with high school level math to systems that can do research level mathematics now and derive novel results in theoretical physics. It's also striking how much progress India has made in its mission to put AI to work for more people in more parts of the country. And India's leadership in sovereign AI
[0:07:16] Building on infrastructure, SLMs, and much more has been great to watch. More than 100 million people in India use ChatGPT every week. More than a third of them are students. India is also the fastest growing market now for Codex, our coding agent that works to help people develop software faster and better. India, the world's largest democracy, is well positioned to lead in AI, not just to build it, but to shape it.
[0:07:43] And decide what our future is going to look like. And it's important to move quickly. On our current trajectory, we believe we may be only a couple of years away from early versions of true superintelligence. If we are right, by the end of 2028, more of the world's intellectual capacity could reside inside of data centers than outside of them. This is an extraordinary statement to make. And of course we could be wrong. But I think it really bears serious consideration.
[0:08:15] A superintelligence at some point on its development curve would be capable of doing a better job being the CEO of a major company than any executive, certainly me, or doing better research than our best scientists. As we prepare for this possibility, we are guided by three core beliefs. Number one, we believe that democratization of AI is the only fair and safe path forward. Democratization of AI is the best way to ensure that humanity flourishes. On the other hand,
[0:08:48] Centralization of this technology in one company or country could lead to ruin. The desirable future a couple of decades from now has got to look like a world of liberty, democracy, widespread flourishing, and an increase in human agency. Some people want effective totalitarianism in exchange for a cure for cancer. I don't think we should accept that trade-off, nor do I need to. AI should extend individual human will. We'll probably need superintelligence to help us figure out the new governance mechanisms.
[0:09:20] To ensure that this happens fairly at scale and to avoid problems like extremely unbalanced compute access or something else. Second, we believe that AI resilience is a core safety strategy. We don't mean that this is the only safety strategy. We will continue to need to build safe systems and solve difficult technical alignment challenges. But increasingly, we need to start broadening how we think about safety to include societal resilience. No AI lab.
[0:09:50] No AI system can deliver a good future on their own. For an obvious example, there'll be extremely capable biomodels available open source that could help people create new pathogens. We need a society-wide approach about how we're going to defend against this. And third, the future of AI is not going to unfold exactly like anyone predicts. And we believe that many people need to have a stake in shaping the outcome. The development of AI has already held many surprises, and I assume there are bigger ones to come.
[0:10:23] We understand that with technology this powerful, people want answers. But it's important to be humble about what we don't know and always remember that sometimes our best guesses are wrong. Most of the important discoveries happen when technology and society meet, sometimes have some friction, and co-evolve. For example, we don't yet know how to think about some superintelligence being aligned with dictators in totalitarian countries. We don't know how to think about countries using AI to fight new kinds of war with each other.
[0:10:54] We don't know how to think about when and whether countries are going to have to think about new forms of social contracts. But we think it's important to have more understanding and society-wide debate before we're all surprised. Of special note, and related to all three points, we continue to believe that iterative deployment is a key strategic insight, and that society needs to contend with and use each successive new level of AI capability, have time to integrate it, understand it, and decide how to move forward.
[0:11:25] This has been working surprisingly well so far. If we are right and systems continue to improve at this pace, it's going to change the economics of a lot of things. A really great thing about AI progress is that it looks like many things are going to get much cheaper and have much faster economic growth. We're already seeing what AI is doing for access to high-quality healthcare, education, and more. In the coming years, we expect to see robots make many products and physical goods cheaper as supply chains get automated.
[0:11:56] The limit to how far this cost reduction can go may only be government policy. But the other side of this coin is that current jobs are going to get disrupted, as AI can do more and more of the things that drive our economy today. It'll be very hard to outwork a GPU in many ways. It'll be easy in some other ways. For example, we really seem hardwired to care about other people much more than we care about machines. We're somewhat less concerned about the long-term future. Technology always disrupts jobs. We always find new and better things to do.
[0:12:30] The people of 500 years ago would have thought that our current jobs often look silly, like ways to entertain ourselves, create stress. And the people 500 years from now hopefully will look at us, hopefully look to us, like impossibly rich people playing games, trying to find ways to pass their times. But we should all hope they feel much more fulfilled than we do today. I'm confident we will keep being driven to be useful to each other, to express our creativity, to gain status, to compete, and much more.
[0:12:59] But the specifics of what we do day to day will probably look very different. Each generation has built on the work of the generations before. And with new tools, the scaffolding gets a little taller. This collective external lattice, the set of tools that we have built up around ourselves, is remarkable. And we are capable of doing things that our great-great-grandparents couldn't have dreamed possible. It is a moral imperative to make sure that our great-great-grandchildren can stay the same. And technology, and especially AI,
[0:13:30] For a democratic AI future, it is not enough to just give people tools and wealth. We also need to give them agency and power. The vision that AI companies lay out fundamentally reduced to either unilateral control or decentralized power. Sharing control means accepting that some things are going to go wrong in exchange for not having one thing go mega wrong. Cemented totalitarian control. This is a fundamental trade-off of democracy, and it is one that we believe in very strongly.
[0:14:04] As the way to give everyone collective agency over the future. Of course, this is not to suggest that we won't need any regulation or safeguards. We obviously do urgently, like we have for other powerful technologies. In particular, we expect the world may need something like the IAEA for international coordination of AI, and especially for it to have the ability to rapidly respond to change in circumstances. The next few years will test global society as this technology continues to improve at a rapid pace.
[0:14:33] We can choose to either empower people or concentrate power. Thank you to be here in India. And it's incredible to see the country's leadership in advanced AI. I was last here a little over a year ago, and it's striking how much progress has happened since then. We've gone from AI systems that struggled with high school level math to systems that can do research level mathematics now and derive novel results in theoretical physics. It's also striking how much progress India has made in its mission to put AI to work for more people.
[0:15:04] in more parts of the country. And India's leadership in sovereign AI, building on infrastructure, SLMs, and much more, has been great to watch. More than 100 million people in India use ChatGPT every week. More than a third of them are students. India is also the fastest growing market now for Codex, our coding agent that works to help people develop software faster and better. India, the world's largest democracy, is well positioned to lead in AI, not just to build it, but to shape it.
[0:15:35] And decide what our future is going to look like. And it's important to move quickly. On our current trajectory, we believe we may be only a couple of years away from early versions of true superintelligence. If we are right, by the end of 2028, more of the world's intellectual capacity could reside inside of data centers than outside of them. This is an extraordinary statement to make. And of course we could be wrong. But I think it really bears serious consideration.
[0:16:07] A superintelligence at some point on its development curve would be capable of doing a better job being the CEO of a major company than any executive, certainly me, or doing better research than our best scientists. As we prepare for this possibility, we are guided by three core beliefs. Number one, we believe that democratization of AI is the only fair and safe path forward. Democratization of AI is the best way to ensure that humanity flourishes. On the other hand,
[0:16:40] Centralization of this technology in one company or country could lead to ruin. The desirable future a couple of decades from now has got to look like a world of liberty, democracy, widespread flourishing, and an increase in human agency. Some people want effective totalitarianism in exchange for a cure for cancer. I don't think we should accept that trade-off, nor do I think we need to. AI should extend individual human will. We'll probably need superintelligence to help us figure out the new governance mechanisms.
[0:17:12] To ensure that this happens fairly at scale and to avoid problems like extremely unbalanced compute access or something else. Second, we believe that AI resilience is a core safety strategy. We don't mean that this is the only safety strategy. We will continue to need to build safe systems and solve difficult technical alignment challenges. But increasingly, we need to start broadening how we think about safety to include societal resilience. No AI lab.
[0:17:42] No AI system can deliver a good future on their own. For an obvious example, there'll be extremely capable biomodels available open source that could help people create new pathogens. We need a society-wide approach about how we're going to defend against this. And third, the future of AI is not going to unfold exactly like anyone predicts. And we believe that many people need to have a stake in shaping the outcome. The development of AI has already held many surprises, and I assume there are bigger ones to come.
[0:18:15] We understand that with technology this powerful, people want answers. But it's important to be humble about what we don't know and always remember that sometimes our best guesses are wrong. Most of the important discoveries happen when technology and society meet, sometimes have some friction, and co-evolve. For example, we don't yet know how to think about some superintelligence being aligned with dictators in totalitarian countries. We don't know how to think about countries using AI to fight new kinds of war with each other.
[0:18:46] We don't know how to think about when and whether countries are going to have to think about new forms of social contracts, but we think it's important to have more understanding and society-wide debate before we're all surprised. Of special note, and related to all three points, we continue to believe that iterative deployment is a key strategic insight, and that society needs to contend with and use each successive new level of AI capability, have time to integrate it, understand it, and decide how to move forward.
[0:19:17] This has been working surprisingly well so far. If we are right and systems continue to improve at this pace, it's going to change the economics of a lot of things. A really great thing about AI progress is that it looks like many things are going to get much cheaper and have much faster economic growth. We're already seeing what AI is doing for access to high-quality healthcare, education, and more. In the coming years, we expect to see robots make many products and physical goods cheaper as supply chains get automated.
[0:19:49] The limit to how far this cost reduction can go may only be government policy. But the other side of this coin is that current jobs are going to get disrupted, as AI can do more and more of the things that drive our economy today. It'll be very hard to outwork a GPU in many ways. It'll be easy in some other ways. For example, we really seem hardwired to care about other people much more than we care about machines. We're somewhat less concerned about the long-term future. Technology always disrupts jobs. We always find new and better things to do.
[0:20:22] The people of 500 years ago would have thought that our current jobs often look silly, like ways to entertain ourselves, create stress. And the people 500 years from now hopefully will look at us, hopefully look to us, like impossibly rich people playing games, trying to find ways to pass their times. But we should all hope they feel much more fulfilled than we do today. I'm confident we will keep being driven to be useful to each other, to express our creativity, to gain status, to compete, and much more.
[0:20:51] But the specifics of what we do day to day will probably look very different. Each generation has built on the work of the generations before. And with new tools, the scaffolding gets a little taller. This collective external lattice, the set of tools that we have built up around ourselves, is remarkable. And we are capable of doing things that our great-great-grandparents couldn't have dreamed possible. It is a moral imperative to make sure that our great-great-grandchildren can stay the same. In technology, and especially AI.
[0:21:23] For a democratic AI future, it is not enough to just give people tools and wealth. We also need to give them agency and power. The vision that AI companies lay out fundamentally reduced to either unilateral control or decentralized power. Sharing control means accepting that some things are going to go wrong in exchange for not having one thing go mega wrong. Cemented totalitarian control. This is a fundamental trade-off of democracy, and it is one that we believe in very strongly.
[0:21:56] As the way to give everyone collective agency over the future. Of course, this is not to suggest that we won't need any regulation or safeguards. We obviously do urgently, like we have for other powerful technologies. In particular, we expect the world may need something like the IAEA for international coordination of AI, and especially for it to have the ability to rapidly respond to change in circumstances. The next few years will test global society as this technology continues to improve at a rapid pace.
[0:22:26] We can choose to either empower people or concentrate power. Thank to be here in India, and it's incredible to see the country's leadership in advanced AI. I was last here a little over a year ago, and it's striking how much progress has happened since then. We've gone from AI systems that struggled with high school level math to systems that can do research level mathematics now and derive novel results in theoretical physics. It's also striking how much progress India has made in its mission to put AI to work for more people.
[0:22:56] in more parts of the country. And India's leadership in sovereign AI, building on infrastructure, SLMs, and much more, has been great to watch. More than 100 million people in India use ChatGPT every week. More than a third of them are students. India is also the fastest growing market now for Codex, our coding agent that works to help people develop software faster and better. India, the world's largest democracy, is well positioned to lead in AI, not just to build it, but to shape it.
[0:23:27] And decide what our future is going to look like. And it's important to move quickly. On our current trajectory, we believe we may be only a couple of years away from early versions of true superintelligence. If we are right, by the end of 2028, more of the world's intellectual capacity could reside inside of data centers than outside of them. This is an extraordinary statement to make. And of course we could be wrong. But I think it really bears serious consideration.
[0:23:59] A superintelligence at some point on its development curve would be capable of doing a better job being the CEO of a major company than any executive, certainly me, or doing better research than our best scientists. As we prepare for this possibility, we are guided by three core beliefs. Number one, we believe that democratization of AI is the only fair and safe path forward. Democratization of AI is the best way to ensure that humanity flourishes. On the other hand,
[0:24:32] Centralization of this technology in one company or country could lead to ruin. The desirable future a couple of decades from now has got to look like a world of liberty, democracy, widespread flourishing, and an increase in human agency. Some people want effective totalitarianism in exchange for a cure for cancer. I don't think we should accept that trade-off, nor do I think we need to. AI should extend individual human will. We'll probably need superintelligence to help us figure out the new governance mechanisms.
[0:25:05] To ensure that this happens fairly at scale and to avoid problems like extremely unbalanced compute access or something else. Second, we believe that AI resilience is a core safety strategy. We don't mean that this is the only safety strategy. We will continue to need to build safe systems and solve difficult technical alignment challenges. But increasingly, we need to start broadening how we think about safety to include societal resilience. No AI lab.
[0:25:35] No AI system can deliver a good future on their own. For an obvious example, there'll be extremely capable biomodels available open source that could help people create new pathogens. We need a society-wide approach about how we're going to defend against this. And third, the future of AI is not going to unfold exactly like anyone predicts. And we believe that many people need to have a stake in shaping the outcome. The development of AI has already held many surprises, and I assume there are bigger ones to come.
[0:26:08] We understand that with technology this powerful, people want answers. But it's important to be humble about what we don't know and always remember that sometimes our best guesses are wrong. Most of the important discoveries happen when technology and society meet, sometimes have some friction, and co-evolve. For example, we don't yet know how to think about some superintelligence being aligned with dictators in totalitarian countries. We don't know how to think about countries using AI to fight new kinds of war with each other.
[0:26:38] We don't know how to think about when and whether countries are going to have to think about new forms of social contracts. But we think it's important to have more understanding in society-wide debate before we're all surprised. Of special note, and related to all three points, we continue to believe that iterative deployment is a key strategic insight, and that society needs to contend with and use each successive new level of AI capability, have time to integrate it, understand it, and decide how to move forward.
[0:27:09] This has been working surprisingly well so far. If we are right and systems continue to improve at this pace, it's going to change the economics of a lot of things. A really great thing about AI progress is that it looks like many things are going to get much cheaper and have much faster economic growth. We're already seeing what AI is doing for access to high-quality healthcare, education, and more. In the coming years, we expect to see robots make many products and physical goods cheaper as supply chains get automated.
[0:27:41] The limit to how far this cost reduction can go may only be government policy. But the other side of this coin is that current jobs are going to get disrupted, as AI can do more and more of the things that drive our economy today. It'll be very hard to outwork a GPU in many ways. It'll be easy in some other ways. For example, we really seem hardwired to care about other people much more than we care about machines. We're somewhat less concerned about the long-term future. Technology always disrupts jobs. We always find new and better things to do.
[0:28:14] The people of 500 years ago would have thought that our current jobs often look silly, like ways to entertain ourselves, create stress. And the people 500 years from now hopefully will look at us, hopefully look to us, like impossibly rich people playing games, trying to find ways to pass their times. But we should all hope they feel much more fulfilled than we do today. I'm confident we will keep being driven to be useful to each other, to express our creativity, to gain status, to compete, and much more.
[0:28:43] But the specifics of what we do day-to-day will probably look very different. Each generation has built on the work of the generations before. And with new tools, the scaffolding gets a little taller. This collective external lattice, the set of tools that we have built up around ourselves, is remarkable. And we are capable of doing things that our great-great-grandparents couldn't have dreamed possible. It is a moral imperative to make sure that our great-great-grandchildren can stay the same. And technology, and especially AI,
[0:29:15] For a democratic AI future, it is not enough to just give people tools and wealth. We also need to give them agency and power. The vision that AI companies lay out fundamentally reduced to either unilateral control or decentralized power. Sharing control means accepting that some things are going to go wrong in exchange for not having one thing go mega wrong. Cemented totalitarian control. This is a fundamental trade-off of democracy, and it is one that we believe in very strongly.
[0:29:48] As the way to give everyone collective agency over the future. Of course, this is not to suggest that we won't need any regulation or safeguards. We obviously do urgently, like we have for other powerful technologies. In particular, we expect the world may need something like the IAEA for international coordination of AI, and especially for it to have the ability to rapidly respond to change in circumstances. The next few years will test global society as this technology continues to improve at a rapid pace.
[0:30:18] We can choose to either empower people or concentrate power. Thank to be here in India, and it's incredible to see the country's leadership in advanced AI. I was last here a little over a year ago, and it's striking how much progress has happened since then. We've gone from AI systems that struggled with high school level math to systems that can do research level mathematics now and derive novel results in theoretical physics. It's also striking how much progress India has made in its mission to put AI to work for more people.
[0:30:48] And more parts of the country. And India's leadership in sovereign AI, building on infrastructure, SLMs, and much more, has been great to watch. More than 100 million people in India use ChatGPT every week. More than a third of them are students. India is also the fastest growing market now for Codex, our coding agent that works to help people develop software faster and better. India, the world's largest democracy, is well positioned to lead in AI, not just to build it, but to shape it.
[0:31:19] And decide what our future is going to look like. And it's important to move quickly. On our current trajectory, we believe we may be only a couple of years away from early versions of true superintelligence. If we are right, by the end of 2028, more of the world's intellectual capacity could reside inside of data centers than outside of them. This is an extraordinary statement to make. And of course we could be wrong. But I think it really bears serious consideration.
[0:31:52] A superintelligence at some point on its development curve would be capable of doing a better job being the CEO of a major company than any executive, certainly me, or doing better research than our best scientists. As we prepare for this possibility, we are guided by three core beliefs. Number one, we believe that democratization of AI is the only fair and safe path forward. Democratization of AI is the best way to ensure that humanity flourishes. On the other hand,
[0:32:24] Centralization of this technology in one company or country could lead to ruin. The desirable future a couple of decades from now has got to look like a world of liberty, democracy, widespread flourishing, and an increase in human agency. Some people want effective totalitarianism in exchange for a cure for cancer. I don't think we should accept that trade-off, nor do I think we need to. AI should extend individual human will. We'll probably need superintelligence to help us figure out the new governance mechanisms.
[0:32:57] To ensure that this happens fairly at scale and to avoid problems like extremely unbalanced compute access or something else. Second, we believe that AI resilience is a core safety strategy. We don't mean that this is the only safety strategy. We will continue to need to build safe systems and solve difficult technical alignment challenges. But increasingly, we need to start broadening how we think about safety to include societal resilience. No AI lab.
[0:33:27] No AI system can deliver a good future on their own. For an obvious example, there'll be extremely capable biomodels available open source that could help people create new pathogens. We need a society-wide approach about how we're going to defend against this. And third, the future of AI is not going to unfold exactly like anyone predicts. And we believe that many people need to have a stake in shaping the outcome. The development of AI has already held many surprises, and I assume there are bigger ones to come.
[0:34:00] We understand that with technology this powerful, people want answers. But it's important to be humble about what we don't know and always remember that sometimes our best guesses are wrong. Most of the important discoveries happen when technology and society meet, sometimes have some friction, and co-evolve. For example, we don't yet know how to think about some superintelligence being aligned with dictators in totalitarian countries. We don't know how to think about countries...