The Changing Physics of Cybersecurity and Drug Development with OpenAI President Greg Brockman

“we think that humans should be firmly in the driver's seat, should have control over the future,”

Claim: The speaker says humans should retain control over the future as AGI is developed.

“and this technology should amplify human agency.”

Claim: The speaker says AI technology should amplify human agency.

“where software in some ways is becoming like a barrier to entry, it's becoming so cheap, you don't just want software, You want software that is reliable, that is secure, that you can believe in,”

Claim: As software becomes very cheap, people will want software that is reliable, secure, and trustworthy, not merely software itself.

“and so cyber spend, security spend should actually vastly outstrip software spend.”

Claim: The speaker argues that cybersecurity spending should greatly exceed software spending.

“I think that people have not yet really internalized how much the... the physics around cyber security are going to change, I think there's an opportunity to end up with a much more secure world.”

Claim: The speaker believes people have not yet grasped how substantially cybersecurity's underlying dynamics will change, creating an opportunity for a much more secure world.

“operates that there's a cycling quote of it never gets easier you just go faster and that's very much what what happens is that there's just so much activity and that every time you think you're at kind of peak sort of you know impact and how things are happening you realize that there's actually another level that's possible and it's not just about any one company it's not just about any one technology it's really about we as a world and society are all coming together to adopt this technology get the most out of it for everyone's lives and so”

Claim: The speaker describes AI progress as continually revealing higher possible levels of impact, driven by society-wide adoption rather than any single company or technology, with the aim of benefiting people's lives.

“I think to me, the math results this week represent a real milestone,”

Claim: The speaker considers OpenAI's math results from that week a significant milestone.

“if you think about like the fundamental driver of human quality of life improving over the past thousand years is technology, It's scientific advancement.”

Claim: The speaker says technology and scientific advancement have fundamentally improved human quality of life over the past thousand years.

“And I think that we now have the ability to massively accelerate scientific discovery. And there's always been this question of.”

Claim: The speaker believes AI now gives people the ability to accelerate scientific discovery substantially.

“Is AI actually able to come up with new knowledge? And at this point, it's just impossible to deny, It is definitively yes possible.”

Claim: The speaker asserts that AI can produce genuinely new knowledge.

“Now there's still more to go. I think we're not yet coming up with whole brand new theories, but we are able to resolve these hundred year old math problems.”

Claim: The speaker says AI has not yet generated wholly new theories, although it can solve mathematical problems that have remained open for about a century.

“you know sort of the just the machine just sort of gives you an answer right it's really what can a human equipped with these tools achieve that was unthinkable before and i think it's almost like the leverage of humans to do whatever you want to do it's going way way up right it's now that you can if you have a math problem you want to go solve you can solve it but what about taking that same capability applying it to biology what about if you're a small business owner and you”

Claim: The speaker argues that AI tools greatly increase human leverage, enabling people to achieve things that were previously unthinkable.

“want to be able to operate your business well like that level of intelligence that's tuned for that domain it's just going to mean that almost this sort of unthinkable results that are aligned with your goals with your desires is now going to be within the reach of everyone and that to me is so exciting”

Claim: The speaker expects domain-specific AI intelligence to help people, including small-business owners, achieve results aligned with their goals and desires, making such capabilities broadly accessible.

“like I came from a math background. Like I thought it was going to be a mathematician.”

Claim: Greg Brockman says he came from a mathematics background and once expected to become a mathematician.

“And I was very inspired by mathematicians like Galois or Gauss, right? These people who were working on like a hundred, 400 year time horizon, right?”

Claim: Greg Brockman says he was inspired by mathematicians such as Galois and Gauss, whom he characterized as working on 100- to 400-year time horizons.

“And I think there's something deeply beautiful about mathematics being about sort of understanding the structure of the universe.”

Claim: Greg Brockman finds it deeply beautiful that mathematics can help people understand the structure of the universe.

“And I think there's something deeply beautiful about mathematics being about sort of understanding the structure of the universe.”

Claim: The speaker views mathematics as a beautiful way of understanding the structure of the universe.

“And in my mind, it's something that the beauty and the meaning of mathematics, it's not something that is objective in some sense. It's in the eye of the beholder.”

Claim: Greg Brockman believes the beauty and meaning of mathematics are not wholly objective and depend on the beholder.

“And in my mind, it's something that the beauty and the meaning of mathematics, it's not something that is objective in some sense. It's in the eye of the beholder.”

Claim: The speaker believes the beauty and meaning of mathematics are subjective rather than wholly objective.

“It's something that appeals to us and there is some amount that's like kind of physical laws and it's about simplicity and, you know, compression theory, you know, it's just complexity theory and all these things like that are maybe are more physics, but there's something that is also importantly human about it.”

Claim: The speaker says mathematics has a human, subjective dimension alongside aspects connected to physical laws, simplicity, compression, and complexity.

“It's something that appeals to us and there is some amount that's like kind of physical laws and it's about simplicity and, you know, compression theory, you know, it's just complexity theory and all these things like that are maybe are more physics, but there's something that is also importantly human about it.”

Claim: Greg Brockman says mathematics has a human dimension because it appeals to people, in addition to aspects connected with physical laws, simplicity, compression, and complexity.

“And so to me, it's not just about, I think a lot of the conversation about mathematics is not just about here's a proof, no one understands it, great, we resolved the problem.”

Claim: Greg Brockman argues that the value of mathematical work lies in its insights, not merely in obtaining a proof that resolves a problem without being understood.

“It's really about what are the insights that are derived? And the thing that I think we will see, but I think there's promising signs of life on, is that there's beautiful insights within the work that these AIs have done.”

Claim: Greg Brockman believes AI systems' work has produced beautiful insights.

“It's really about what are the insights that are derived? And the thing that I think we will see, but I think there's promising signs of life on, is that there's beautiful insights within the work that these AIs have done.”

Claim: The speaker sees promising signs that AI-generated mathematical work contains beautiful insights.

“You know, we published about almost 400. very important open problems that are now resolved and the mathematical community is going through all that to look for insights and if you read some of the things that people have posted you know domain experts in these particular problems and they say that it's like that there's these almost like sort of very surprising ways of applying ideas or incredibly like wait why do they use that result over here and that that what's coming out is not these incomprehensible”

Claim: Greg Brockman says nearly 400 important open mathematical problems have been resolved and the mathematical community is reviewing the results for insights.

“You know, we published about almost 400. very important open problems that are now resolved and the mathematical community is going through all that to look for insights and if you read some of the things that people have posted you know domain experts in these particular problems and they say that it's like that there's these almost like sort of very surprising ways of applying ideas or incredibly like wait why do they use that result over here and that that what's coming out is not these incomprehensible”

Claim: The speaker says nearly 400 important open mathematical problems have been resolved and the mathematical community is reviewing the solutions for insights.

“very important open problems that are now resolved and the mathematical community is going through all that to look for insights and if you read some of the things that people have posted you know domain experts in these particular problems and they say that it's like that there's these almost like sort of very surprising ways of applying ideas or incredibly like wait why do they use that result over here and that that what's coming out is not these incomprehensible”

Claim: Greg Brockman says domain experts reviewing AI-produced mathematical solutions have reported surprising applications of ideas and unexpected uses of results.

“important points and one of maybe the underappreciated benefits that I expect from and hope for and we're starting to see from this technology is helping people know more understand more understand each other better understand the world better and I think that there are technical mechanisms that can underlie the”

Claim: Greg Brockman expects and hopes that AI technology will increasingly help people gain knowledge, understand one another, and understand the world, and says early signs of this benefit are appearing.

“We have an AI that's able to go and solve these problems, see all these connections between domains, but how does it actually communicate it to you? How do you understand it?”

Claim: Greg Brockman says AI can solve problems and identify cross-domain connections, while emphasizing the challenge of communicating its results in ways people understand.

“And actually there's even connections to AI safety and alignment on having AI that we're able to understand what they're doing.”

Claim: Greg Brockman connects AI safety and alignment with making AI systems' actions understandable to people.

“There are ways to train the AI that are actually very aligned with having humans be able to have oversight, to be able to be in the driver's seat, which is a fundamental value, a fundamental thing that is important for the future from OpenAI's perspective,”

Claim: Greg Brockman says a fundamental OpenAI value for the future is ensuring humans can oversee AI and remain in control.

“but also are very much aligned with what you're saying of can you make the results of an AI be something that means you learn something that is really tuned for the human rather than just sort of knowledge for some abstract knowledge sake.”

Claim: Greg Brockman argues that AI outputs should be designed to help people learn, rather than provide knowledge merely for its own sake.

“we've had a number of years and iterations of this technology to see the shape of what to expect.”

Claim: Greg Brockman says that several years of AI development and iterations have revealed the general shape of what to expect from the technology.

“capable in a certain dimension and if you understand the like why it got capable in that dimension then all other fields of human endeavor that have similar properties you should also expect similar results and so in my mind like why is mathematics something that we've managed to really transform in the way that we have had such incredible results and that part of it is that our AI is now at a certain level of capability and raw intelligence”

Claim: Greg Brockman argues that when AI capability in one dimension is understood, similar results should be expected in other fields of human endeavor that share the relevant properties.

“capable in a certain dimension and if you understand the like why it got capable in that dimension then all other fields of human endeavor that have similar properties you should also expect similar results and so in my mind like why is mathematics something that we've managed to really transform in the way that we have had such incredible results and that part of it is that our AI is now at a certain level of capability and raw intelligence”

Claim: Greg Brockman says that AI's transformation of mathematics is partly due to the models reaching a certain level of capability and raw intelligence.

“sense part of it is that you could think of it as that well mathematics is checkable but that's not quite true anymore right like we really started with very checkable problems of here's an arithmetic problem there's like some numeric answer and so you have an AI that you can train to just like check if it got the answer right because we can very much perceive is the answer that the AI is proposing the same as the actual answer but for a proof very different proofs are very hard to check is this actually accurate or not formalized proof okay yes that is checkable but but natural language proof which is what we're really producing here much much more difficult but the thing that actually i think is the core insight is really about data right it's that you can get lots of data in mathematics you get lots of data in theoretical physics can you get lots of data in biology that's obviously something very data bottlenecked very difficult now fortunately we also have been working lots of techniques to be able to work with less data”

Claim: Greg Brockman says that arithmetic answers are readily verifiable, whereas proofs, particularly natural-language proofs, are much harder to verify.

“actually accurate or not formalized proof okay yes that is checkable but but natural language proof which is what we're really producing here much much more difficult but the thing that actually i think is the core insight is really about data right it's that you can get lots of data in mathematics you get lots of data in theoretical physics can you get lots of data in biology that's obviously something very data bottlenecked very difficult now fortunately we also have been working lots of techniques to be able to work with less data”

Claim: Greg Brockman says mathematics and theoretical physics offer abundant data, while biology is data-bottlenecked and its data is difficult to obtain.

“actually accurate or not formalized proof okay yes that is checkable but but natural language proof which is what we're really producing here much much more difficult but the thing that actually i think is the core insight is really about data right it's that you can get lots of data in mathematics you get lots of data in theoretical physics can you get lots of data in biology that's obviously something very data bottlenecked very difficult now fortunately we also have been working lots of techniques to be able to work with less data”

Claim: Greg Brockman says that researchers have developed many techniques for working with less data.

“and this is the fundamental physics of deep learning, right? Compute data algorithms equals capability equals results.”

Claim: Greg Brockman characterizes deep learning as a relationship in which compute, data, and algorithms yield capability and results.

“We're able to do it in these hard domains where people have spent hundreds of years of our smartest people all trying to solve the problem and now the AI can solve it. Amazing.”

Claim: Greg Brockman says AI can solve some difficult problems that the most capable human researchers have worked on for hundreds of years.

“When you transfer that over to biology and clearly what we want is drug discovery, all of these things that are like fundamental changes in biology and our understanding of it to improve quality of life, that's what we want to apply.”

Claim: Greg Brockman says AI should be applied to biology to advance drug discovery and biological understanding in ways that improve quality of life.

“The question will be, and it is not yet resolved, do we have the data, do we have the ability to get the data?”

Claim: Greg Brockman says it remains unresolved whether sufficient biological data exists or can be obtained.

“And so I think that what will happen is that we will take these, and we're already doing this, where we take these models, these same techniques, and apply them in the biological domain.”

Claim: Greg Brockman says OpenAI is already applying the same AI models and techniques to biology.

“sort of case for clinical trial and all these things and it's just like you can write those things with an AI much more easily and so I think that we're going to have to look at every step of the drug discovery process and there are going to be places where the fundamental data just does not exist and so then innovations on wet labs innovation on how you actually do simulations all of this will become incredibly important but so I think that”

Claim: Greg Brockman predicts that the drug-discovery process will need to be examined at every step, including steps where the necessary underlying data does not exist.

“sort of case for clinical trial and all these things and it's just like you can write those things with an AI much more easily and so I think that we're going to have to look at every step of the drug discovery process and there are going to be places where the fundamental data just does not exist and so then innovations on wet labs innovation on how you actually do simulations all of this will become incredibly important but so I think that”

Claim: Greg Brockman predicts that wet-lab innovation and improved simulation methods will become extremely important for drug discovery where fundamental data is missing.

“sort of case for clinical trial and all these things and it's just like you can write those things with an AI much more easily and so I think that we're going to have to look at every step of the drug discovery process and there are going to be places where the fundamental data just does not exist and so then innovations on wet labs innovation on how you actually do simulations all of this will become incredibly important but so I think that”

Claim: Greg Brockman predicts that AI will make writing clinical-trial case materials and related documents much easier.

“We will see these breakthroughs coming,”

Claim: Greg Brockman predicts that breakthroughs in applying AI to biology will occur.

“but it's important to know there are things that are shared with mathematics, there's about scientific discovery, and things that are very different in terms of where the data comes from.”

Claim: Greg Brockman says AI applications to biology share some aspects of mathematical discovery but differ in the sources of their data.

“Look, I think that drug discovery is the most clear and obvious area if you get to pick one that you want to fully revolutionize and just in places where you're going to really uplift everyone's quality of life that is just like so purely important and maybe the top one application of AI that everyone's been kind of thinking about for a long time”

Claim: Greg Brockman considers drug discovery the clearest candidate for a full AI-driven revolution because of its potential to improve quality of life broadly.

“clearly got to do it and to your point like there's this whole pipeline it's just so long and I think we're going to find that humans are incredibly bad at biology incredibly bad at drug discovery and if you think about it there's this like”

Claim: Greg Brockman says humans are very poor at biology and drug discovery.

“But for a neural net, I don't really think there's any reason to believe it should be better in natural language than it should be better, you know, compared to biology.”

Claim: Greg Brockman says there is no particular reason to expect neural networks to perform better in natural language than in biology.

“And so the fact we're seeing such incredible results with neural nets and natural language, you're just like, from first principles, you should be able to really be able to crack the language of biology.”

Claim: Greg Brockman argues that neural networks' strong results in natural language suggest, from first principles, that they should be able to understand biological systems.

“And so to me, that is number one, it's like first principles that should work in terms of what we're seeing,”

Claim: Greg Brockman says biology is the top AI application that should work according to first principles and the results currently being observed.

“we're so close on so many axes.”

Claim: Greg Brockman says AI is close to achieving breakthroughs across many dimensions.

“It can massively increase the parallelism. I think it can decrease so many of these steps in terms of their duration.”

Claim: AI can substantially increase parallelism and shorten the duration of many drug-discovery steps.

“There's even like lots of people who the drug to cure their particular condition exists. It's just that it's just out there somewhere in the literature that there's a paper and if you can find that one paper.”

Claim: For some conditions, a potentially curative drug already exists, but the relevant evidence is buried in the literature and must be found.

“super is relevant to your conditions then you'll be able to cure the person and there's plenty of precedent for it and it can massively scale up those kinds of operations i think that impact in the short run is actually very possible but material science another example and i think there's a long tail of other domains where we should just see real changes in in kind of what is possible for humanity to accomplish”

Claim: The speaker says there is substantial precedent for identifying literature relevant to a person's condition and using it to cure that person.

“super is relevant to your conditions then you'll be able to cure the person and there's plenty of precedent for it and it can massively scale up those kinds of operations i think that impact in the short run is actually very possible but material science another example and i think there's a long tail of other domains where we should just see real changes in in kind of what is possible for humanity to accomplish”

Claim: The speaker considers it very possible in the short run that AI will greatly scale operations for finding treatments relevant to individual conditions.

“super is relevant to your conditions then you'll be able to cure the person and there's plenty of precedent for it and it can massively scale up those kinds of operations i think that impact in the short run is actually very possible but material science another example and i think there's a long tail of other domains where we should just see real changes in in kind of what is possible for humanity to accomplish”

Claim: The speaker expects AI to produce substantial changes in what humanity can accomplish in materials science and many other domains.

“Part of it is you're doing like successively more expensive trials that have a, you know, different set of safety considerations resulting in large scale human trials.”

Claim: Drug development involves progressively more expensive trials with differing safety considerations, culminating in large-scale human trials.

“And then these trials today are, they're ordained by regulation, right, for good reason.”

Claim: Clinical trials are regulated, and the speaker says there are good reasons for that regulation.

“It feels like one of the limiting factors if you believe the technology will work and parts of the data can be acquired or generated is going to be. the convincing regulators that we should change the expectations around what drugs are accepted and what the steps are.”

Claim: If the technology works and relevant data can be acquired or generated, persuading regulators to change drug-acceptance expectations and required steps may be a limiting factor.

“I think there is a moment here where, and there's more work to be done to reach this point, but if you can really prove that the physics of drug discovery have changed, then of course it should trickle through to every part of the system.”

Claim: If evidence establishes that the underlying science of drug discovery has changed, the speaker believes every part of the system should adapt.

“about self-driving cars right we have so much of our how roads operate and safety and all these things that are really tuned for human drivers and if you have if you reach a point where self-driving cars are 10 times or 100 times safer than that you've reached that point where AI is actually able to deliver the results in a domain that just are so unambiguously the thing that you want, of course you should then say, OK, how do we rethink the whole transportation system?”

Claim: If self-driving cars become 10 to 100 times safer than human-driven cars, the speaker believes the transportation system should be reconsidered.

“about self-driving cars right we have so much of our how roads operate and safety and all these things that are really tuned for human drivers and if you have if you reach a point where self-driving cars are 10 times or 100 times safer than”

Claim: Road operations and safety practices have been designed around human drivers.

“Maybe you should have regulation that says you only are allowed to have self-driving cars in certain areas, like there are a lot of potential sort of changes to the way that you think about.”

Claim: The speaker suggests that regulation could limit self-driving cars to designated areas.

“what is even trying to come out of a system that needs to trickle back into regulation as well. And of course, regulation is not uniform across the world, right, that there's very different regulations in Australia versus the US when it comes to clinical trials.”

Claim: Clinical-trial regulations differ between Australia and the United States.

“And they're already seeing a lot of people moving their trials to places where the regulatory environment makes more sense for how things are, how they want to pursue and what people think is efficacious for them.”

Claim: Some people are already relocating clinical trials to jurisdictions whose regulatory environments better match their preferred approach and views of efficacy.

“exactly how things should change I think will depend a lot on what we actually can prove right is it the case you actually reach the point that you can get better results from a neural net than from actually running a clinical trial like we're not there but”

Claim: The speaker says decisions about changing drug-discovery practices should depend substantially on what evidence can establish.

“if they were more predictive on efficacy or tox or something, then obviously you should change it.”

Claim: If neural networks become more predictive of drug efficacy or toxicity, the speaker believes drug-development practices should change.

“what is possible has fundamentally changed and let's very quickly then reorient around it and again you have to do it you have to really think through all the all the consequences right and all of the implications and that one of the you know if you're thinking about long-term studies people and my question of like you know is this neural net actually in terms of what's predicting the stimulation actually accurate for long-term outcomes it's pretty hard to not have at least had you know one long time tick”

Claim: The speaker argues that society should quickly adapt when capabilities fundamentally change, while carefully evaluating the consequences and implications.

“what is possible has fundamentally changed and let's very quickly then reorient around it and again you have to do it you have to really think through all the all the consequences right and all of the implications and that one of the you know if you're thinking about long-term studies people and my question of like you know is this neural net actually in terms of what's predicting the stimulation actually accurate for long-term outcomes it's pretty hard to not have at least had you know one long time tick”

Claim: The speaker says evaluating a neural network's predictions of long-term outcomes is difficult without at least one long-term observation.

“But I think we will see like 80% of what is possible come to fruition very quickly.”

Claim: The speaker predicts that 80% of what is possible will be realized very quickly.

“I think DOTS has really changed how many of us work.”

Claim: The speaker says DOTS has substantially changed the way many people work.

“is a proactive agent it's always on always working you stop working it keeps working and the kind of thing that my dots does right so I have a dot that's looking out for work that I was supposed to do like message I was supposed to send things I said I would do that I haven't gotten to yet and I always wake up in the morning there's like a list of things that I was like oh yeah this is my to-do list”

Claim: DOTS is a proactive agent that continues working when its user is not working.

“is a proactive agent it's always on always working you stop working it keeps working and the kind of thing that my dots does right so I have a dot that's looking out for work that I was supposed to do like message I was supposed to send things I said I would do that I haven't gotten to yet and I always wake up in the morning there's like a list of things that I was like oh yeah this is my to-do list”

Claim: The speaker's DOTS agent tracks unfinished intended tasks, including messages to send and commitments to fulfill, and surfaces them as a morning to-do list.

“best and that it often notices things that I never even ask it to notice if there's like a conflict on my calendar in terms of scheduling it'll point that out if it's like hey you actually need to leave this meeting earlier in order to get to the other thing and so it's like so contextually aware it knows what I'm doing with my day it understands the projects we're working on it understands I sort of you know it knows what's in my inbox knows what's in my slack and is able to integrate across these and it knows the questions I've been asking about so it knows what”

Claim: The speaker's DOTS agent can proactively identify unrequested issues, including calendar conflicts and schedule adjustments needed to get to another meeting.

“best and that it often notices things that I never even ask it to notice if there's like a conflict on my calendar in terms of scheduling it'll point that out if it's like hey you actually need to leave this meeting earlier in order to get to the other thing and so it's like so contextually aware it knows what I'm doing with my day it understands the projects we're working on it understands I sort of you know it knows what's in my inbox knows what's in my slack and is able to integrate across these and it knows the questions I've been asking about so it knows what what areas are most of top of mind for me and I think that having like in some ways what dots is is a change to the form factor not the capability right it's like it's an astral level model it's got its own cloud computer but that the change in that form factor in a workplace setting with this level of intelligence it's a game changer”

Claim: The speaker says DOTS integrates context about his schedule, projects, inbox, Slack, and current questions or priorities.

“what areas are most of top of mind for me and I think that having like in some ways what dots is is a change to the form factor not the capability right it's like it's an astral level model it's got its own cloud computer but that the change in that form factor in a workplace setting with this level of intelligence it's a game changer”

Claim: The speaker characterizes DOTS as a workplace form-factor change rather than a capability change, combining an Astral-level model with its own cloud computer, and calls this combination a game changer.

“got one. I've got one.”

Claim: The speaker says he has one DOTS agent.

“It's just sort of doing everything across my work life.”

Claim: The speaker says his DOTS agent operates across his work life.

“There is some, I think these things are going to converge, no question.”

Claim: Workplace assistance and consumer assistance are likely to converge.

“And I think that there are some pressures that are very different in different environments. So for a mass market consumer product, really figuring out how to scale and to provide things for free or at very low cost is very important. Whereas in the workplace,”

Claim: Mass-market consumer AI products need to scale and offer services for free or at very low cost.

“productivity if you're the executive individual contributor whatever your role is then it is much easier to say peak intelligence and really as long as you're getting returns to paying more it's worthwhile to throw more computer at it and so I think that that is the main distinction is really around cost and we've seen this in chat GPT right that chat GPT we put in a lot of effort to have the largest scale free tier and bringing things like memory”

Claim: The main distinction between consumer and workplace personal assistants is the cost and compute each market segment can support.

“productivity if you're the executive individual contributor whatever your role is then it is much easier to say peak intelligence and really as long as you're getting returns to paying more it's worthwhile to throw more computer at it and so I think that that is the main distinction is really around cost and we've seen this in chat GPT right that chat GPT we put in a lot of effort to have the largest scale free tier and bringing things like memory”

Claim: OpenAI invested substantial engineering effort in offering a large-scale ChatGPT free tier and making memory available to free users.

“productivity if you're the executive individual contributor whatever your role is then it is much easier to say peak intelligence and really as long as you're getting returns to paying more it's worthwhile to throw more computer at it and so I think that that is the main distinction is really around cost and we've seen this in chat GPT right that chat GPT we put in a lot of effort to have the largest scale free tier and bringing things like memory”

Claim: For workplace productivity, higher intelligence and additional compute can be justified when the resulting returns warrant the higher cost.

“to all of our free users was an incredible endeavor like a real huge engineering effort and it's just very easy to implement the same functionality at much higher cost for a very small number of people and so I think that what we're seeing in in the personal assistant workspace is that there is there are going to be these different market segments where different amounts of compute are going to be expended fundamentally and so sometimes when there's different level of compute it then means the product”

Claim: Personal-assistant products will use different amounts of compute across market segments.

“to all of our free users was an incredible endeavor like a real huge engineering effort and it's just very easy to implement the same functionality at much higher cost for a very small number of people and so I think that what we're seeing in in the personal assistant workspace is that there is there are going to be these different market segments where different amounts of compute are going to be expended fundamentally and so sometimes when there's different level of compute it then means the product”

Claim: The same assistant functionality can be provided to a small number of users at much higher cost than a mass-market free offering.

“product can look a little different right that it can be that there's just more functionality that's embedded in one versus the other and there's of course different use cases you think about things like”

Claim: Differences in compute allocation across assistant products can result in differences in their embedded functionality.

“you know payments and travel and things like that are much more for the consumer but i don't think there's any fundamental difference right it's like really just like an ai that's doing things and it's just like happens to be like slight sort of pressure in one way or another and so one of the strengths of open ai is the fact that we have 1.2 billion consumers on chat gbt and we have also a leading coding and enterprise business and no one else is both in house and so that a lot of our strategies to say let's unify these by these so we're not sort of you know sort of competing on two fronts it's really competing on one front and i think that this is also the general direction of travel of ai is to blur the distinction between what is consumer what is enterprise we're going to have this renaissance of entrepreneurs like we're already seeing it i mean i imagine you're seeing this very concretely um”

Claim: OpenAI aims to unify its consumer and enterprise strategies so it competes on one front rather than two.

“you know payments and travel and things like that are much more for the consumer but i don't think there's any fundamental difference right it's like really just like an ai that's doing things and it's just like happens to be like slight sort of pressure in one way or another and so one of the strengths of open ai is the fact that we have 1.2 billion consumers on chat gbt and we have also a leading coding and enterprise business and no one else is both in house and so that a lot of our strategies to say let's unify these”

Claim: Consumer and workplace AI do not differ fundamentally; they are AI systems doing tasks under somewhat different pressures.

“you know payments and travel and things like that are much more for the consumer but i don't think there's any fundamental difference right it's like really just like an ai that's doing things and it's just like happens to be like slight sort of pressure in one way or another and so one of the strengths of open ai is the fact that we have 1.2 billion consumers on chat gbt and we have also a leading coding and enterprise business and no one else is both in house and so that a lot of our strategies to say let's unify these”

Claim: Payments and travel are use cases more associated with consumer assistants than workplace assistants.

“you know payments and travel and things like that are much more for the consumer but i don't think there's any fundamental difference right it's like really just like an ai that's doing things and it's just like happens to be like slight sort of pressure in one way or another and so one of the strengths of open ai is the fact that we have 1.2 billion consumers on chat gbt and we have also a leading coding and enterprise business and no one else is both in house and so that a lot of our strategies to say let's unify these”

Claim: OpenAI's strength is its combination of 1.2 billion ChatGPT consumers with leading coding and enterprise businesses; Brockman says no other company has both in-house.

“by these so we're not sort of you know sort of competing on two fronts it's really competing on one front and i think that this is also the general direction of travel of ai is to blur the distinction between what is consumer what is enterprise we're going to have this renaissance of entrepreneurs like we're already seeing it i mean i imagine you're seeing this very concretely um”

Claim: The general direction of AI is toward blurring the distinction between consumer and enterprise products.

“by these so we're not sort of you know sort of competing on two fronts it's really competing on one front and i think that this is also the general direction of travel of ai is to blur the distinction between what is consumer what is enterprise we're going to have this renaissance of entrepreneurs like we're already seeing it i mean i imagine you're seeing this very concretely um”

Claim: AI is expected to enable a renaissance of entrepreneurship, which Brockman says is already underway.

“I know people who are quitting their job to go start companies who say I never would have done it if I didn't have an AI that I knew I could rely on that could take care of so much of the normal barriers to entry.”

Claim: Some aspiring founders say they are leaving jobs to start companies because they can rely on AI to take care of many ordinary barriers to entry.

“And so I think that we're going to see just like the distinctions have held for so long, I think they're going to change.”

Claim: The longstanding distinctions between consumer and enterprise are expected to change.

“So first of all, I think that the best UI is no UI. And I think in AI, that is where we're going.”

Claim: Brockman believes AI interfaces are moving toward a best-case experience with no conventional user interface.

“It's just one of the things you can talk to, and you can speak to it verbally, you can talk to it over Slack, iMessage, any of the surface you can call it, and that's literally what you can do with dots, like literally today.”

Claim: As of the conversation, dots can be accessed through verbal interaction, Slack, iMessage, or other surfaces.

“an admin right you have kind of your company security team you have just more stakeholders and it's not just sort of a ai that's for you that's representing you that you own and you own its outcomes and so i think that that is the main distinction i see between the enterprise and the in the home use case and of course this leads to product differences as well and sometimes there are very small things like for example when we release a new feature”

Claim: Brockman sees the main distinction between home and enterprise AI as personal users owning an assistant that represents them and its outcomes, unlike workplace deployments with multiple stakeholders.

“an admin right you have kind of your company security team you have just more stakeholders and it's not just sort of a ai that's for you that's representing you that you own and you own its outcomes and so i think that that is the main distinction i see between the enterprise and the in the home use case and of course this leads to product differences as well and sometimes there are very small things like for example when we release a new feature”

Claim: Workplace AI deployments involve administrators, company security teams, and more stakeholders than personal use.

“that enterprises want control over when it's introduced and so we have admin controls and toggles and so you can have an experience where there's just less functionality available to you.”

Claim: Enterprise controls can result in some functionality being unavailable to workplace users.

“that enterprises want control over when it's introduced and so we have admin controls and toggles and so you can have an experience where there's just less functionality available to you.”

Claim: Enterprise customers want to control the timing of new feature releases, and OpenAI provides administrative controls and toggles for that purpose.

“as a result of the fact that it is in the workplace and that does then change the well in some way to surface that to you and this comes back to well the whole point was that there was no UI right so it's like what does it mean for there to be functionality that's missing and how do you even surface it and let people know that that's the case so that's an ongoing challenge and I think that that will not magically go away I think that you need to lean into that that complexity but yeah to me fundamentally it's kind of crazy and beautiful if you think about it the chat GBT”

Claim: In AI products with little or no UI, explaining that functionality is unavailable remains an ongoing challenge that must be addressed rather than expected to disappear.

“first of all, I do think the physics of cybersecurity are changing,”

Claim: The speaker believes the underlying dynamics of cybersecurity are changing.

“and we're in what we call the defender's window.”

Claim: The speaker characterizes the current period as a “defender’s window.”

“And the way that I think about this window is that we have peak frontier AI capabilities that can help defenders secure themselves.”

Claim: During the defender’s window, peak frontier AI capabilities are available to help defenders secure their systems.

“They're not broadly diffused yet,”

Claim: Peak frontier AI capabilities that can help cybersecurity defenders have not yet become broadly diffused.

“but the broadly diffused capabilities are ratcheting. And you can see this from each model release to model release from providers across the world, the capabilities on cyber that are...”

Claim: Cyber capabilities in broadly available AI models are increasing with successive model releases from providers around the world.

“available to everyone are going up and you're starting to see the leading edges of some of the impacts right you're starting to see for example in South Korea that there were some high-profile hacks of banks that were driven by different models and I think that the picture that is so critical is that defenders need to use this time and so we have been working very hard to be able to responsibly give access”

Claim: The speaker says different AI models drove several high-profile bank hacks in South Korea.

“available to everyone are going up and you're starting to see the leading edges of some of the impacts right you're starting to see for example in South Korea that there were some high-profile hacks of banks that were driven by different models and I think that the picture that is so critical is that defenders need to use this time and so we have been working very hard to be able to responsibly give access”

Claim: Cyber capabilities available to the public are increasing, and some leading-edge effects are beginning to emerge.

“available to everyone are going up and you're starting to see the leading edges of some of the impacts right you're starting to see for example in South Korea that there were some high-profile hacks of banks that were driven by different models and I think that the picture that is so critical is that defenders need to use this time and so we have been working very hard to be able to responsibly give access”

Claim: Cybersecurity defenders need to take advantage of the current period to strengthen their defenses.

“available to everyone are going up and you're starting to see the leading edges of some of the impacts right you're starting to see for example in South Korea that there were some high-profile hacks of banks that were driven by different models and I think that the picture that is so critical is that defenders need to use this time and so we have been working very hard to be able to responsibly give access through our Daybreak program to defenders to be able to use our peak intelligence or to use very capable models to find and fix vulnerabilities in their systems. And one thing that is very important to realize is these flaws”

Claim: The company has worked to responsibly provide defenders access through its Daybreak program to highly capable models for finding and fixing system vulnerabilities.

“have existed, right? It's it's every line of human written code in some ways that you have to ask, is this secure?”

Claim: The software vulnerabilities under discussion predate the current AI-related cybersecurity concerns.

“Was the developer trained in the right way? And you just start to realise that the way in which we built systems were kind of fundamentally insecure and we always knew it, right?”

Claim: The speaker says systems were built in fundamentally insecure ways, a problem people already recognized.

“There's never been a CISO who's been over resourced, right?”

Claim: The speaker asserts that CISOs have never had more resources than they needed.

“exactly and so i think that to me and i think very importantly it's very important to realize these are really human issues right these are the way that we chose to build the internet it was security got kind of bolted on right and we had high profile incidents right you think about early worms and viruses and then people started to realize well we need”

Claim: The speaker characterizes cybersecurity problems as human-created issues arising from internet design choices that treated security as an add-on.

“virus scanning software, And we need to have encryption for requests, you know, over HTTP should be over HTTPS. All of these mechanisms were ones where I think that we kind of took the slower route in order to add security on afterwards.”

Claim: The speaker says mechanisms such as virus scanning and encrypted HTTPS requests were introduced as later additions to improve security.

“Like the internet's really been built for resilience, not necessarily for security.”

Claim: The internet was designed primarily for resilience rather than security.

“And I think that what we're going to see now is a moment where A lot of that debt, a lot of that tech debt is going to come to the surface. It's time for us to pay it down.”

Claim: The speaker expects accumulated internet technology debt to become more apparent and believes it is time to address it.

“And as a company, we are doing, we are really leaning into trying to help with this moment.”

Claim: The speaker says the company is actively trying to help address the current cybersecurity transition.

“So for example, we have a billion dollar commitment that we've made to help frontline defenders get access to our models and utilize them for defense that we're.”

Claim: The company has committed one billion dollars to help frontline defenders access and use its models for cybersecurity defense.

“studying and trying to find different vulnerabilities out there and then thinking about how do we help the world respond how do we coordinate incident response how do we work with government how do we work with the community in order to land these as well as possible and to help people navigate through but I think that people have not yet really internalized how much the physics around cyber security are going to change I think we will end up I think there's an opportunity to end up with a much more secure world”

Claim: The speaker believes the current changes create an opportunity for the world to become much more secure.

“studying and trying to find different vulnerabilities out there and then thinking about how do we help the world respond how do we coordinate incident response how do we work with government how do we work with the community in order to land these as well as possible and to help people navigate through but I think that people have not yet really internalized how much the physics around cyber security are going to change I think we will end up I think there's an opportunity to end up with a much more secure world”

Claim: The company is studying vulnerabilities and planning how to support responses, coordinate incident response, and collaborate with governments and the community.

“studying and trying to find different vulnerabilities out there and then thinking about how do we help the world respond how do we coordinate incident response how do we work with government how do we work with the community in order to land these as well as possible and to help people navigate through but I think that people have not yet really internalized how much the physics around cyber security are going to change I think we will end up I think there's an opportunity to end up with a much more secure world”

Claim: The speaker believes people have not yet fully grasped the extent to which cybersecurity’s underlying dynamics will change.

“You think about the fact that AI lets you pay down your tech debt in a way that was just previously unthinkable.”

Claim: AI enables organizations to reduce technical debt in ways that were previously unthinkable.

“The fact that you could do things like formal verification of software, like these models are obviously great at proving things. Why not prove security properties of software?”

Claim: The speaker argues that AI models’ ability to prove things could be applied to verifying software security properties.

“That's something that the technology has kind of been there for people, but it's just so much work.”

Claim: Formal verification technology has been available, but applying it has required substantial work.

“And so I think all of that can change. And I think the world we're moving to right now, and we've done this internally within OpenAI, is every time a new model drops, you need to apply it to secure yourself, find your vulnerabilities.”

Claim: Applying each new AI model to an organization’s infrastructure to find vulnerabilities can change how organizations approach security.

“is and then saturate and so for example we took about 25% of our production engineers and said all of your projects are on hold you are just securing now and took our peak intelligence model at the time had it poke at all of our infrastructure raising a bunch of things built some tools so that it's easy to do it”

Claim: OpenAI temporarily reassigned about 25% of its production engineers to security work and used its most capable model at the time to search its infrastructure for vulnerabilities.

“again over time and we found a number of issues we fixed it and at some point we saturated the issues that this model was capable of finding and that's a very good story right it's like i think that the thing to realize is that we're gonna have a next model we're gonna have to repeat that process and that that will continue but at some point again i think if you think about well the code that's being written”

Claim: OpenAI will need to repeat its security-review process as new models become available.

“again over time and we found a number of issues we fixed it and at some point we saturated the issues that this model was capable of finding and that's a very good story right it's like i think that the thing to realize is that we're gonna have a next model we're gonna have to repeat that process and that that will continue but at some point again i think if you think about well the code that's being written”

Claim: In OpenAI’s security exercise, the model found and helped address issues until the organization reached the limit of issues that model could identify.

“So much of it is now AI-assisted. We can actually train these models, and we are training these models, to write secure code from the beginning.”

Claim: Much of the code being written is AI-assisted, and OpenAI is training AI models to produce secure code from the outset.

“And so I think we can have sort of fundamentally more secure systems through the use and applications of these AI models in the right way.”

Claim: Using AI models appropriately can make systems fundamentally more secure.

“Yeah, look, I I have I have a lot of optimism about how we can navigate this moment. I think it does require a coordinated response,”

Claim: The speaker is optimistic that society can navigate the current cybersecurity moment, provided there is a coordinated response.

“It really requires defenders to internalize how the physics of cybersecurity are changing, that there is this window where that window is not going to last forever,”

Claim: Cybersecurity defenders need to understand the changing dynamics of cybersecurity and act during a window of opportunity that will not last indefinitely.

“And that you should think about we're shifting to this world where the security of our organization is related to. both the fundamentals that they've gotten right which is kind of one-time investment but also the level of cyber capability they've used to defend themselves and as we shift to that world I think that we can be in a world where defenders are able to outstrip attackers but it is going to take a lot of of change right it's going to take a sort of change in terms of how we think about every organization operating and how we collectively think about critical infrastructure and resourcing them”

Claim: An organization’s security depends on both sound foundational practices and the cyber capabilities it deploys to defend itself.

“both the fundamentals that they've gotten right which is kind of one-time investment but also the level of cyber capability they've used to defend themselves and as we shift to that world I think that we can be in a world where defenders are able to outstrip attackers but it is going to take a lot of of change right it's going to take a sort of change in terms of how we think about every organization operating and how we collectively think about critical infrastructure and resourcing them those that have sort of not had those resources to date.”

Claim: Defenders may be able to outstrip attackers, but doing so will require major changes to organizational practices and collective support and resourcing for critical infrastructure.

“exactly it's becoming so cheap and so of course making sure you don't want software right you want software that is reliable that is secure that you can believe in and so to me the resilience of that software is the sort of that's exactly that's going to become even more important so I think if you have this moment we have this opportunity to you know to your point of the XKCD with the little block at the bottom of some open source maintainer like like that never made any sense Mm and”

Claim: As software becomes increasingly cheap, the value of software that is reliable, secure, and trustworthy will increase.

“was because of some of this physics right of kind of how hard it is to write software and to write reliable software and that everyone has the supply chain and so few companies audit their supply chain right if there's we've seen this with supply chain vulnerabilities where someone uploads a malicious package and a number of companies get hit all of that needs to change and it was totally impossible to change with human software developers because there's so few or there's so few people who have this expertise”

Claim: Because reliable software is difficult to produce and relevant expertise is scarce, broad software-supply-chain auditing was very difficult to achieve.

“We're partnering with companies like CrowdStrike, and I think that it is, again, a coordinated response,”

Claim: OpenAI is partnering with companies such as CrowdStrike as part of a coordinated cybersecurity response.

“but we really view it as a core priority for OpenAI to lead, because we have the frontier model, we have the peak intelligence model, certainly in cybersecurity,”

Claim: OpenAI considers leading cybersecurity a core priority because it believes it has the frontier model and the peak-intelligence model in cybersecurity.

“we think that there is transformative potential for the good.”

Claim: OpenAI believes its models could have transformative positive effects in cybersecurity.

“possible but that means that we as a company and we've been going through a security transition right that over the past couple months we've talked about safety security alignment of our development process but we're also thinking about that for the world of how to really help the world become more fundamentally secure and help people through this whole period”

Claim: OpenAI is considering how to help improve security worldwide and support people through the current period.

“possible but that means that we as a company and we've been going through a security transition right that over the past couple months we've talked about safety security alignment of our development process but we're also thinking about that for the world of how to really help the world become more fundamentally secure and help people through this whole period”

Claim: Over the past couple of months, OpenAI has been transitioning its development process to address safety, security, and alignment.

“I think that ethos is very core to our mission, To ensure AGI benefits all of humanity.”

Claim: OpenAI's mission is to ensure that artificial general intelligence benefits all of humanity.

“We think that humans should be firmly in the driver's seat, should have control over the future.”

Claim: OpenAI believes humans should remain firmly in control of AI and retain control over the future.

“These AIs are more capable than anyone today, right? Even GPT-4 knows far more than any person.”

Claim: The speaker says current AI systems are more capable than any person today and that GPT-4 knows far more than any individual.

“And I think that the same spirit needs to apply to AI, the fact of the spirit of humans are in the driver's seat,”

Claim: Human control should remain a guiding principle in the design and use of AI.

“like human judgment human values and maybe you should think of how humans will operate in the in certain fields in certain dimensions in the future whatever it is is become more like CEOs or like board members but it's important that we have that level of agency and that I think that”

Claim: In some future domains, humans may act more like CEOs or board members while retaining agency through human judgment and values.

“We're going to build machines just like you know if you're a CEO you're clearly going to have domain experts you're clearly going to have teams they're going to have ape knowledge across so many things and maybe even better ideas than you on many different axis but fundamentally the question is are you happy is this what you wanted is this what makes your life better and that that to me is something that we should all collectively it's not just about AI it's about technology technology is to lift up human civilization it's the single most fundamental driver of improvement in quality of life and that is what we should be seeking from AI that is the belief of open AI thank”

Claim: OpenAI believes AI should improve quality of life and contribute to human civilization, as technology's fundamental purpose is to lift up civilization.