Name: OpenAI CTO Mira Murati responds to Elon Musk calling Apple partnership ‘creepy spyware’
Uploaded: 2024-06-12
Duration: 21 min 4 s
Description: Murati sat down with editor-at-large Michal Lev-Ram at a Fortune MPW dinner in San Francisco to discuss the Apple partnership, new CFO Sarah Friar, customer trust and safety, how she found her love for AI, and more.

Murati sat down with editor-at-large Michal Lev-Ram at a Fortune MPW dinner in San Francisco to discuss the Apple partnership, new CFO Sarah Friar, customer trust and safety, how she found her love for AI, and more.

OpenAI CTO Mira Murati responds to Elon Musk calling Apple partnership ‘creepy spyware’

Tech

Murati sat down with editor-at-large Michal Lev-Ram at a Fortune MPW dinner in San Francisco to discuss the Apple partnership, and more.

Mira, thank you for being here with us. I know you must be a little bit busy. Um So I have like a gazillion questions for Mira, but we have 20 minutes. So I just wanna set everybody's expectations. We're going to stick to a few topics including um some uh recent news and some areas that Mira oversees as, as CTO. Um And hopefully we can dive into some of those. Um I think my, my first question for you actually is just given how insanely busy and given the barrage of news that, you know, some good, some bad. Um You joined around six years ago, I think um at that time, it was a very different organization. You were relatively speaking, kind of, you know, under the uh off the radar. Um Do you ever miss those days of being able to be head down and doing the work? I mean, we're still very much heads down doing the work. It's just that the work has evolved. Uh And it's not just about the research is also now that, you know, research has progressed so far, it's also about figuring out how do we bring this into the world in a way that's helpful and safe. Um So the mission remains the same, but we made a lot of progress on research and the work has extended. Um And yes, there is a lot of public attention on it as well. Um, which can feel unusual when you're working on technology and building products. Um But, you know, it's also very much necessary given the importance of what we're doing and is good. OK. Um Let's go back even further. Um For those who don't know your trajectory, um give us kind of the, the, the, the highlights I know you grew up in Albania at a time when even uh decent access to the internet wasn't a given. How did your interest in A I start? And what do we need to know about? What led you to your current role? I've always been interested in math and physics. And so my, my background uh is very much math heavy. And uh I went on to study engineering work in aerospace and then in automotive at Tesla. And that sort of, that's where I started to get more and more interested in applications of A I in the real world through self-driving cars. Um And then I went on to apply A I and computer vision and um um applications of virtual reality and augmented reality sort of trying to understand how do you create a new interface where you interact with information in a more natural way more intuitive. Um And that's, that's sort of where I was pushing the applications of A I in the real world in different domains. Um And then, you know, thought about how do I, how do I learn a lot about A I? And um at the time I was considering, you know, um applications of A I in different domains and thought ultimately what I wanted to do the most was learn about general artificial intelligence and really diving deep into research. Um And then from there understanding what, what we could do with it in the real world. Um And I was really attracted to open mission, uh which is, you know, at the time opening, I was a nonprofit and now we're a cap profit, but the mission is still the same. Uh If the company B and OK, um I wanna get into some of the, the latest news and again, there's, there's a lot. Um but um I know Maram mentioned Sarah Fryar. Um you hired on just yesterday, the company announced that you have a new um chief product officer and um CFO. And um I think probably a lot of the women in this room are familiar with Sarah. Um She is a seasoned uh public company executive, both, you know, former CEO and CFO. And I feel like it's kind of like when you get married and the next day someone asks you, when are you gonna have kids? Uh this is like you hire a seasoned publicly public company executive. And the next day people ask you, does it, what should we be reading into this as an IP O in the, in the midst, what, how should we be uh deciphering these hires? What does it mean for the company? What it really means is that we're, you know, in this next phase of the company where we're serving hundreds millions of users and developers and companies out there. And, you know, we need to have the most competent and best executives, helping us to grow in this next phase of the company. And we're very excited about Sarah and Kevin and the skills and uh the leadership they bring into the company. OK. Um I had a feeling you wouldn't say anything about the IP O. Um So I want to talk about um the news with Apple. Um I'm guessing everybody saw this or you wouldn't live in this area. Um um But so I help us like unpack a little bit. Um you know, from a technology perspective, just how differentiated and significant this is. Um I know Apple and this is Apple, not open is, but Apple referred to it as the most advanced security architecture ever deployed for cloud A I compute at scale. Um First of all, do you, do you agree with that for this integration um with Apple Intelligence and again, just help us understand what's going on behind the scenes that theoretically makes it so private and secure. Um Well, so the partnership with Apple is we're super excited about it because it's such a great opportunity for us to bring our technology to as many people out there as possible in a very seamless way where, you know, they don't have to switch devices, they can interact with, you know, the technology comes to you in the best way possible and safest way possible. Um And the partnership with Apple is very aligned, we care very deeply about privacy and safety of our products. Um and how we develop the technology and how we ultimately deploy it. So that's very much um we're very much in alignment and it's, it's going to push us forward in a better direction. Um So Apple obviously is, you know, has been known for uh taking a privacy and, and, you know, secure approach um to roll out. Do you think that for people who have anxiety about interacting with gen A I products, not just chat GP T, do you think that this kind of helps uh uh you know, push in that general direction of more trust? Ultimately, we have to do a lot to build um trust around A I in general, right? There are so many aspects to it. Um And I think this partnership is uh hopefully, you know, something that, that moves the needle in the right direction as we bring A I into so many products and different integrations. And um specifically with our partnership, you know, people can log into charge GP T or through their Apple account. And we have, um, you know, the, the product policies that apply to either Cha GP T or Apple. Um But, you know, we both strongly believe in privacy and safety of the products. Um We will not be logging information through the um Apple accounts. Um But you know, when people log in on CHA GP T, then we will be, we will be, and, and, and you know, they have not specifically opted out from us being able to uh look at the data. Then in that case, I in in those cases, we will um we will be looking at the data and when people have specifically opted out, then we will not be looking at the data more broadly. We do not train on uh any customer data in general um or you know, other or, or data where people have, have specifically opted out. And by the way, does, on that note, the fact that the models can't train on the da user data um of of Apple os users. Um I mean is that a downside here uh to pushing forward the ability for these models to keep training and that's a lot of data. Um I'm look the the reality is that we have to do the right thing for the users. And it's very important that people are trusting the technologies and the products in which they're deployed. And it's important for people to have control over their data, how it's used and very, very important for them to understand how it's used and where it's used. Um So that comes first and that's a priority regardless of the impact that it has in how, how much, how much it drives the technology forward or, and all. So you're gonna have to forgive me because I'm gonna quote Elon Musk. Um I'm sorry, but uh but he is one person that has, is, is not uh in agreement with this being a model for privacy and security. He called the integration creepy spyware. What's your response to that? Huh. Don't care. That's, that's his opinion. I, I mean, I'm not, yeah, I uh obviously, I don't think so. We don't think so. Uh And we care deeply about the privacy of our users and about the safety of our products and how we're developing and deploy this technology. Um Our track record in deploying the technology safely is very strong. We've done that with G BT 34. We're doing the same thing with 40, right now, and we're trying to be as transparent as possible with the public and communicate how we're making decisions with a preparedness framework that shares how we think about frontier models and our approach to safety there. Um And we will continue to be transparent through the preparedness framework through things like the spec where I don't think anyone else has such a developed and transparent framework for what happens with model behavior. Um It's, you know, from the high level rules all the way to very technical decisions and how complex they are. Uh We're sharing that with, with the public and we're providing an opportunity for people to provide input. And we're also working on um a lot of experimental ways to gather input from non users as well. Um I think these are the important things, the work is not done, we have a lot more to do here. Um And not just opening eye, I think the whole industry in sort of explaining what's going on with the technology, um creating more participation, more agency. But we have to start with actually explaining because the biggest risk is that stakeholders um misunderstand the technology. So in the, in the midst of all of um these developments and you know, current partnerships um that, that you guys are striking, um you are also currently training a new model, a new A I model which is expected to bring the company closer to A G I. And I know you spend a great deal of time thinking about this. Um And, and also thinking through worst case scenarios, best case scenarios, like I'm curious, um you know, and I think probably everybody in this room is very curious about what that future looks like. Can you talk to us a little bit about how you um think about it. Um how you scenario plan for it and what's like the the the mirror case scenario like what's the ideal view of this future in, in your view? Um So my framework for thinking about the progress that we're making uh and where this technology is going is that it's definitely going to be incredibly transformative technology. Um And it requires a lot of work, not just from the companies that are developing and deploying these technologies, but also from uh other stakeholders like civil society, media, um governments regulators, the general public people in every domain. You know, we, we have to create some sort of shared responsibility around the general preparedness for bringing this technology into the, into the world. But in order to have shared responsibility, we need to understand it and we need to make it very accessible. Um And so we've been focusing on this, this last two parts, making it accessible. Um engaging with people educating with educating various stakeholders on what what the technology means, what it's capable of. Uh The, the most recent thing we did was the release with the release of GB D 40 which is our only model. Um We made it accessible to everyone for free. And you know, I I don't think I don't think there is um enough emphasis on, on how unique that is for the stage where the technology is today in the sense that, you know, inside the labs, we have these capable models and, you know, they're not that far ahead from what the public has access to for free. And that's a completely different trajectory for bringing technology into the world that what we've seen historically. And it's a great opportunity because it brings people along. It gives them uh intuitive sense for the capabilities and risks and allows people to prepare for the advent of this, of, of bringing advanced A I into the world. And obviously, the opportunities are huge. Now, it's, it's normal that we talk a lot about the risks uh because they're so important. Uh But also we wouldn't do this if we didn't believe that the opportunities here are huge. I personally think that there are incredible. Um There is just this incredible promise with advanced A I and health care and education. We're just kind of touching on some of these capabilities and opportunities with personalized tutors like from Khan Academy. But you can imagine that, you know, anyone in the world, in the most remote areas, they would have access to custom tutoring that is, you know, personalized to the way they think to their um cultural background, to their interests. And that's just amazing. Like we normally don't really think about how we learn or how we think until, you know, 10 years of education or so on. And it's just such a missed opportunity. And I really do believe that if you can push human knowledge forward, you can push society and civilization forward. Um And I mean, there's so many opportunities in health care, um, thinking about solutions in climate change. There's a long list of things that we haven't tackled. And I think, uh bringing this advanced A I systems in the world will help us a lot. Um And obviously, you know, there's, there are always risks whenever you have something so powerful, you have to deal with, with the downsides with the misuse. Um And if you think about the world around us, you know, it's all engineering and engineering carries risks and we have learned to, you know, build systems and build um build trust and uh to figure out how to, how to make engineering safe around us, you know, the buildings, the roads, the bridges, everything. Um And so this is quite similar where we have to build all this infrastructure that we trust and makes the deployment of these technologies safe and useful. Um And it's actually very congruent with business needs because people want products that they can trust and that are safe. So, and I wanna ask you one more question just on the safety side, um this is moving so fast, faster than bridge building and, you know, uh real life infrastructure. Um I know you oversee safety, the safety teams. Um and one of the things that uh when you were on our cover last year, we talked about was your belief that the unique structure at open A I provides the necessary guardrails to make sure that you have safety in mind. Obviously, there's been a lot of scrutiny, there was a lot of upheaval late last year. It's hard to remember that you were interim CEO for uh for a little while there. Um because so much had has happened, tell us uh why and how uh you think that you believe that the right guard rails, the right structure on the safety side is still in place. Um I assume you believe that or you wouldn't be there at this point. Well, so uh I, I think you're referring to the board structure, right? OK. Um So the way that opening I was set up was, you know, the question has always been what is the best administrative structure, the best incentive structure, alignment and design? So the company can fulfill its mission in the best possible way. And there have been a couple of iterations, you know, as, as we, as we work towards the mission and we learn more. It is a process of discovery. I mean, we're literally in the pursuit of discovery, we're doing research and with that, you have to kind of learn and adjust and make changes. So there was the initial change from nonprofit to cap profit um in order, you know, to actually be able to fund the supercomputers and hire talent while, you know, putting the mission first. And so that is the goal always to put the mission first. Um But as we discovered in November, you know, with a nonprofit board structure that we had, it doesn't put um the, the non profit board structure didn't have accountability to anyone but themselves. Um And we're building and deploying the technology that will affect everyone. So, ideally, there would be, you know, some sort of accountability on, on, on the board as well. And uh our, our current board uh which is very ex experience in nonprofit and uh larger companies. They are doing a lot of thinking on the governance structure and what is the best way that you can balance the incentives and put the mission first while having a lot of accountability? Ok. Well, as I said, we will not get to uh uh you know, a tiny fraction even of the questions I'd love to ask you, but I really, really appreciate you taking time um and, and sitting here and just being so thoughtful. Um There's so much interest in what you guys are doing and we appreciate you coming here and talking to us and thank you so much with that. We'll let everybody eat. Um Thank you so much. Thank you, Mia.