OpenAI's Fidji Simo on why ads are coming to ChatGPT

14:25Ellis:Starting a business can seem like a daunting task, unless you have a partner like Shopify. They have the tools you need to start and grow your business. From designing a website, to marketing, to selling, and beyond, Shopify can help with everything you need. There's a reason millions of companies like Mattel, Heinz, and Allbirds continue to trust and use them. With Shopify on your side, turn your big business idea into... Sign up for your$1 per month trial at shopify.com slash special offer. Last night, you spent two hours deciding what to wear to the party. This morning, it'll take you two minutes to list it on Depop and make your money back.

15:01Ellis:Just grab your phone, snap a few photos, and we'll take care of the rest. The sheer dress and platform heels you'll never wear again? There's a birthday girl searching for them right now. Your one-and-done look is about to pay for your next night out. Or at least the ride home. Your style can make you cash. Start selling on Depop, where taste recognizes taste.

15:28Fidji Simo:there was someone who i follow who does marketing and anthropic who posted about this and i thought you might find this interesting she said the brief was ambitious one introduce claude to help people understand what's at stake in your choices and if we've accomplished even one of those goals that's meaningful progress maybe they got the second one helping people understand what's at stake. I don't know if this ad actually introduced Claude at all to anyone.

15:56Ellis:Yeah, I mean, it's a good clear binary with the values, but I'm just not sure the public believes that tech is capable of actually standing by any of these claims at this point, when so many companies have given in to stuff that they said they would never do over the years. I also just don't think the level of awareness for Claude is very high. And it's certainly a choice to spend your ad talking about what you're not or about someone else. As you heard in the last episode, I thought the ad was really funny and the performances are like super on the nose of exactly what people are feeling with AI these days and why it feels sycophantic, as they say.

16:39Ellis:But yeah, I might have to put the hashtag only time will tell stamp on this one.

16:46Fidji Simo:What about OpenAI's ad, the Codex ad? People were catching some good Easter eggs. They were giving a shout out to their secretive robotics unit in the video. So they are training these robotic arms that they want to use to automate their AI data centers is what Sam has been saying, and then eventually do humanoids. And the way they've been training them is to see if the arms can put a duck in a cup. So they showed that in the video. And then people were also reading into the POV of the commercial being seemingly through glasses and suggesting like, oh, does that mean they're working on glasses with Johnny Ive?

17:25Fidji Simo:There was even a freeze frame that the official ChachiBT account posted with blurry, glassy looking things being held in front of the humanoid robotic arms. So yeah, what did you make of this Codex ad?

17:37Ellis:Yeah, I thought it was a fairly classic kind of tried and true formula for an ad, which is showing a whole diversity of people building things from scratch and then talking about how we're going to help you do it. And so, yeah, very much, I would say, unsurprising, but but direct Super Bowl ad. I mean, I saw someone tweeting about like how, what was it? They were saying they were asking a bunch of family and friends who aren't in tech what they thought of the different Anthropic and OpenAI ads. And they're like, yeah, the OpenAI ad was kind of inspiring. Like, yeah, go get them. Seize the day.

18:13Ellis:All right. That's the goal. And they're like, and the Claude one, I didn't really even know what it was talking about. So it is tough. I mean, with an audience of 150 million people, we've talked about this on the show before. Or how could you ever know all of them or how to talk to all of them or what the normies really feel about anything without doing the research that we don't usually see? So much of our messaging that these companies are usually doing is just for each other, billboards in the valley. And that is a very different muscle. And if there is something that I've learned in advertising over the years, it's that simple, direct, inspiring, like the OpenAI ad is how you do move the needle with perception.

18:56Ellis:and you just repeat that shit over and over again. Whereas trying to make a sophisticated or nuanced take is not always a good bet. The Facebook chair ad was not a good bet. Oh my God. I would like to know how many people remember that beautiful gem of a moment. For those who don't know, was that a Super Bowl ad or was that just - I think it was. I mean, it certainly felt like one. People called it Facebook as a chair, because that is what the ad felt like it was. And it was showing all these kind of like random scenes of these gathering places or simple utilities in daily life, one of which being a chair.

19:40Ellis:And it talks about Facebook being the chair. And I think the kind of intellectual 4D chess here is talking about yourself as a fundamental utility in the future of the world. It's so simple and people don't even think about it. But yeah, at the end of the day, all people think is, wait, it's a chair?

20:00Fidji Simo:Quickly though, there was also some weird OpenAI fake ads. You saw a fake Alexander Skarzard ad where it was supposedly the device with Johnny Ive, the orb, and some fancy looking earphones. It looked very well produced. And then there was like a Reddit post that was claiming to be a former OpenAI employee saying how frustrated they were that this ad never got out and they're leaking it. And then OpenAI said it was fake news. There was also a fake headline about OpenAI changing its Super Bowl ad at the last minute to respond to Anthropic right before the Super Bowl. So a ton of just FUD and fake news swirling around OpenAI right now.

20:40Ellis:I mean, I am a gullible person. Anyone I know can tell you that. But I I thought it checked all the boxes for what I would have expected them to do. And that's why I thought it was real. I mean, it's funny. I think what made me think it was fake was the Reddit post, the disgruntled employee. But look what it does. It has a current understated superstar, probably a bit overdone, which I think tech executives usually fall for. They want to do someone who's currently cool, even if they're in everything. it had a square aspect ratio almost like a Yorgos Lanthimos ad which he also did another ad for Squarespace it was very mysterious and interesting and almost like 2001 monolith of Alexander Skarsgård tapping this little stainless steel orb what's it going to be it's a mystery we don't know what it is yet and I also thought that given the leaks this fit the bill of what we had heard Johnny Johnny and Sam were developing.

21:38Ellis:I mean, at Apple, we did. Everything's kind of a bit translucent white. I think it makes sense that if they want to do something universal and not Apple-y, that it is stainless steel. Doesn't matter how many fingerprints it picks up. That's just what Johnny wants to do. It also had the open ear design that I felt like we had seen. And I think it all felt very much like I would not have been surprised if this was the tease for the new product until it said dime at the end. And what was the line that came right after that? I forget. But yeah, Dime is a horrible name. Yeah, I mean, even if it was fake, I was nonetheless very impressed for this faux highbrow tech ad that is pretty much exactly what I would have expected had it been real.

22:23Fidji Simo:Unfortunately not. So your favorite commercial, though, is Google's, right? You love the Gemini ad.

22:28Ellis:Oh, yes. Thank you for reminding me. Google wins again just by doing the classic heartfelt story thing. Pulled on your dad heartstrings, didn't it? Obviously. I mean, the tech set is so often trying to come up with productivity use cases, which makes sense. We all love to make things that make us faster and make our brains bigger. But what they did was they told a generative AI story of a kid who is apprehensive about moving into his new house. And his parents use Gemini to create visions of the future house with all his stuff in it. his room ordered just how he wants, but in his new room. The garden with all the things that they haven't planted yet, imagining what they could look like.

23:12Ellis:And they've been doing this for 20 years now, going back to the Parisian love one and others have tried to copy it since. And it's just tried, true, heartstrings. Doesn't matter that that's not everybody's use case. It's just a good Super Bowl ad.

23:29Fidji Simo:I liked it. I liked that it showed the products too. I mean, going back to Anthropic and even OpenAI, which I guess did show Codex on the screen a little bit, but the product was super integrated into the ad. So you knew exactly what you were seeing, which I liked. And also Google really has tuned the way it's messaging these things because it's been doing a lot of Gemini helping people with their family ads.

23:50Ellis:A new kind of help from Google.

23:52Fidji Simo:And it's had to like pull ads because some of them have been so cringe. I remember one where the dad was trying to use Gemini to like write his kid a birthday card or something like that. And the messaging was very off because it was like AI is replacing his intentionality with his daughter. Whereas this is like AI is part of the intentionality. It's part of the relationship, which I appreciated that tuning. It seems like Google has had to take a few swings to get this right, but they finally are.

24:21Ellis:Yeah. It's interesting too. I think the folks at Apple, They don't really do on-screen product storytelling with software in this way, but the tries that they had with AI, I mean, gosh, they're a year and a half old at this point. The ones with Bella Ramsey or otherwise of like the mother who forgets a gift. And so she makes an AI memory movie and presents it as a lot of hard work and authentic. Like that's useful, but not really cool or emotional. Whereas Google just kind of nails it again. And I don't know if they're working with the same agency or what after all these years. I know some of the folks who've kind of been involved in that type of work here or there.

25:01Ellis:But it's really, it's a template at this point. And all you have to do is just come up with the heartfelt story and let it be.

25:08Fidji Simo:Ellis, quickly before we get to this interview with Fiji, I had a really cool experience that I've been dying to talk with you about. Is that so? Yes. So last week, I teased this on the last episode. I was in San Francisco doing a highly secretive briefing with Johnny Ive and the new electric Ferrari. So I spent the day with Johnny, Mark Newsom, his co-founder at Love From, a bunch of other Love From people, pretty much the entire Ferrari leadership team. And they showed us the interior of the all new electric Ferrari Luce, as it's called. And I do not normally do car junkets. This is actually my first one.

25:47Fidji Simo:I don't cover cars, but you know, the Johnny of this all made it very intriguing to me. And I've gotten to see Johnny a little bit since he left Apple, but this was by far the most time I've, I've spent with him. And I'm going to be teasing out kind of all this in the interview and in sources, it'll be out by the time this episode drops. But yeah, man, it was a very cool experience. Have you gotten to see the video of the interior yet?

26:11Ellis:I did. I did. It's a whole new design language for them. I think the biggest question for me was for all the hullabaloo about bringing back buttons, they're still kind of presented as part of the screen, which I think on the one hand makes sense. You know, you kind of have this big dashboard, part of it is a screen, part of it is buttons. I like that they still kept, you know, things like the speedometer. But it distinctly feels a bit less muscly, I think, for a reason. I mean, that's one of the fun things with cars, I think, is all the dials and knobs and this and that. And when you look at it on that screen, and the performance mode looks like a widget on your Mac, I feel like I wonder if that could have been better as as another dial or a knob or something that you could rev, you know, I'm saying in a circular motion, I mean, not as big as the things Mini Cooper has done over the years.

27:04Ellis:But yeah, I was interested in kind of the way that they tried to bridge the gap between all screen and all buttons.

27:13Fidji Simo:Yeah, it's interesting that you feel like it's still not analog enough because it's definitely a way more analog car than any luxury electric vehicle on the market or most new luxury, even gas vehicles on the market. So the entire interior is aluminum and glass. And Johnny was saying it's the most detailed CNC work that he's ever seen or that anyone he's ever worked with has ever done. So that's obviously saying he thinks this is better CNC work than an Apple device. And it should be for what it's going to cost. I'm sure this thing's going to cost like half a million dollars. They haven't revealed the price yet.

27:49Fidji Simo:They were just showing us the interior. I actually didn't even get to see the exterior of the car yet. They deconstructed this like an Apple event where you go in and there's a bunch of parts laid out on Apple-like tables in the different colors. You've got everything like floating in the middle of the room, deconstructed, and where you can walk around it and look at it. And man, it's beautiful. It is definitely a different direction. And I hear you that maybe there are areas where it could be even more analog. But I was frankly just blown away with how tactile everything was. I mean, you can really tell the level of thought that they put into every button click, every knob, every switch.

28:32Fidji Simo:And yeah, definitely a new kind of car.

28:35Ellis:Yeah, I think a lot of it, to be clear, I know I have to be a critic on a podcast. A lot of it looks absolutely stunning. I mean, the launch button, the big Ferrari button you can press like in the center console. I love the instrument cluster behind the steering wheel. There's like a fucking aeronautical style G-forces thing, like just absolutely badass. But you know what this reminds me of, which is in Love From DNA, one of their designers who I believe is still there, Mike Mattis, is he still there? Did you get to talk to him? Oh yeah, he worked on this. Because, okay, see, it reminded me of Nest, the circular interfaces and some of the stuff here.

29:17Ellis:And I'm like, this looks like Mike's DNA. Yeah.

29:20Fidji Simo:So glad you brought that up. There was a moment where we were talking with Johnny and Mark Newsome about the temperature control because there's no auto setting for the temperature control. And apparently that was a big debate they had of you just set the temperature and that there's no auto button. Johnny was talking about the way that the temperature information moves. And he was like, yeah, we had Mike work on this. So it's, you know, that's like only a thing that something like Love From could do is you have like the original designer of Nest work on the Ferrari's temperature control.

29:48Ellis:I'm a huge fan of Mike and everything he's done. For those who don't know, he did a lot of the original iPhone applications. He did Nest. He's done some other startups in between. He also did Facebook Paper, one of my favorite apps ever. And yeah, I felt like with these, I mean, Nest is just so goaded for so many reasons in terms of what it didn't do. And the mixture of, you know, screens and turnable rings that still retains the thermostats familiar, you know, muscle memory to it. Yeah. So this was super interesting to see that kind of hybrid between the Ferrari design language and theirs. Yeah.

30:27Fidji Simo:I mean, the key is made of ink that changes color and lights up when you put it in the vehicle. The dials you were talking about the instrument cluster. So they're physical dials, but they're digital screens. So it's like a four panel series of digital screens with physical instruments integrated throughout. They kept the parallax effect. So when you look to the side, you see the parallax. Oh man. You know, the Apple car for me kind of hung over this whole briefing because he obviously worked on the car when he was at Apple. It was a project that was going on for like a decade. Apple's never confirmed its existence.

31:02Fidji Simo:They reportedly killed it at one point a few years ago. And so I think me and a few of the other tech, more tech minded reporters who were there were kind of looking for strains of, is this the spiritual successor to what would have been the Apple car? And Johnny basically said no, but it's still hard to not see that Apple, you know, eye and that Apple taste in the products in the interior.

31:28Ellis:I mean, we're never going to know about what the car might look like, but I think between CarPlay and this, we probably have a damn good sense of what the actual driver controls may have looked like. Obviously, the prioritization is going to be a little bit different for Ferrari versus what would have been Apple. But I mean, if you look at the little switches that they have on the control panel, you know, for the temperature and this and that, there is a tactile piece, but then right above it, there is a little screen seemingly behind that panel that is dynamic and shows you. So you do get like a nice kind of combination of the tactile and the power of screens to be dynamic.

32:03Ellis:I do think, you know, you would have seen probably something similar. I mean, you see with Johnny's quotes about this, like they wouldn't have made the whole thing a touchscreen because of exactly what Johnny said, and he's right, is that when you're using a touchscreen, you're taking your eyes off the road. And I mean, obviously, you know, we could we could argue about, well, in the future, you're not going to have to look at the road.

32:24Fidji Simo:Yeah. But anyway, very cool experience with Johnny Ive. And with that, I guess let's finally kick it to our interview with OpenAI's CEO of Applications, Fiji SEMO.

32:42Ellis:This episode is brought to you by Indeed. Stop waiting around for the perfect candidate. Instead, use Indeed Sponsored Jobs to find the right people with the right skills fast. It's a simple way to make sure your listing is the first candidate C. According to Indeed data. Sponsored jobs have four times more applicants than non-sponsored jobs. So go build your dream team today with Indeed. Get a $75 sponsored job credit at Indeed.com slash podcast. Terms and conditions apply. In a classroom of sodas, most stay quiet. Then there's Mr. Pibb. Sweet cherry, bold outbursts, the kind of flavor that gets attention.

33:21Bold kick of cherry.

33:23Ellis:Hey yo, Mr. Pibb.

33:29Fidji Simo:Fiji, thank you so much for having us here at your lovely home. Thank you so much for having me on the podcast. It's good to be here. There's so much to talk about. This week, I feel like in particular, we're talking Friday right before the Super Bowl. There was a lot. I was on like three press calls with the Okinawa team this week. You guys are shipping a lot. I want to start with the ads, though. So Anthropic released its Super Bowl ads and Sam has responded. You've seen them, right? Yes, I have. You've seen them. We've all seen them. Sam has responded. Kate has responded, your CMO. But I'd love to know what you thought when you saw those ads.

34:07Yeah. Well, look, the ads are funny. So there's that. At the same time, you know, the thing that I felt most acutely is like, we're at this very interesting moment in time where we can really remind everyone that they're a creator and with AI you can build anything. Like that's really the story of the entire industry. And that's the story of our Super Bowl ad. And I find it that, you know, everyone in the industry could make the pie bigger for everyone if we can make everyone realize that AI is here to give you these superpowers. And I think that's such a much more powerful message than just hitting at a competitor, you know, but they do them and we're going to do us.

34:48It gets people talking. It gets people talking for sure.

34:51Ellis:As a wannabe marketing theorist, I feel like the thing that actually is going to stick out more to people in the Super Bowl, I think they're going to forget about the ads message completely because most people haven't even experienced it yet inside of any of these services.

35:05Fidji Simo:That was the zinger that Sam had was there were more Chachapeteers in Texas than Claude has globally.

35:10Ellis:But I think what people are going to remember, which is interesting culturally, is just the impeccable actor performances that portrayed this very like, you know, we say sycophantic, confident, happy, robotic. Obviously, you guys know better than anybody else that that stuff is going to be fine tuned out and worked through over time and people can choose the tone of voice. But that is what I think is going to make people perk their ears up and laugh. And yeah, if anything, I think that might actually overshadow the ads message outside of tech. And, you know, on the ads message itself, like, I think that's going to be a very interesting debate, right?

35:51On like, is AI truly for everyone or only for the people who can pay for it? And I think I want to live in a world where we give people the choice. Like, you have the choice to pay if you can. And you have the choice to, you know, see ads if you don't want to pay. But it's really important that we find a business model that allows us to distribute the best levels of intelligence to everyone. And I think if we don't introduce ads and, you know, we end up like, you know, portraying ads as like this evil thing, we're going to end up in a much, much worse place in terms of AI diffusion and like the promise of AI, which is empowering everyone.

36:29So, you know, I think it's a bit of a missed opportunity to actually set the industry in the place that it needs to be set. And as for the turn, yes, I think it's a moment in time. It's funny. And I think we're continuing to iterate on the chat GPT personality so that everyone finds the personality they want. But that's also the thing, right? Like everyone wants something different. When I talk to our users, there are some users who love the like super cheerful, like, you know, personalities that pump them up. And then you have some users who are like, no, no, I want a one liner. And like the art is going to be like giving everyone the tools to personalize for what they want.

37:06Fidji Simo:As a OG Facebooker, I think you'll appreciate this. This week has got me thinking a lot about the Facebook versus Apple ATT tracking era where there were the barbs were flying in kind of a similar way between two companies. And they were attacking each other's business models with kind of similar arguments. And it's interesting to see now that that's kind of come to AI. And I wonder, like, as someone who's been in tech for as long as you have, when you see that happening again, what does that say about where AI is right now? I mean, I think that's exactly what you imagine, which is we're at the point where diffusion and scale is the most important thing.

37:47And it turns out we have concluded as an industry over and over again, ads is a pretty good business model to keep products free. and it has created an immense amount of value in tech, right? And for all users. And so, you know, we're at the point where maybe other companies that have very small user bases don't need to think about that. But for us, when you reach, you know, 800 million people and you're like, I want to give as much intelligence as possible to everyone, of course, ads is, you know, a consideration. Now, the thing I think is really important is to realize that ads has also evolved, right?

38:24Like we are not going to do ads the way other companies in the past have done ads because we have a fundamentally different medium. So we came out when we announced ads with super strong principles like ads are not going to influence the response that IGPT gives you because we think full independence from ads and trust that like the trust that people have in these responses is absolutely critical. So we're not going to like copy paste models of the past. We're not going to, you know, obviously we produce all of the bad that some ads platform have done. We want to create a very different kind of ad platform that is very appropriate for this moment, this product.

39:00And AI is different.

39:02Ellis:I think covering ads and even working within an ads business at Snapchat for such a long time, I feel like there's often a tendency for founders, even the ones I work with today, to make big, lofty, noble promises. And then when times get a little tougher, it's very easy to turn that dial on the ad load or this or that. How do you hope to protect against that internally as you guys navigate different market conditions, competitors, etc.? Because is that not the easiest button to press when things get tough as a business? So, you know, the thing that's interesting about us compared to fully ad supported companies that we have many different business models, right?

39:41We have a consumer subscription business model. We have ads. We have B2B companies buying our services.

39:48Fidji Simo:You'll have devices with Johnny in the future. Exactly, exactly. And so that means we have many different revenue streams, and that allows us not to have to max out any single one of them. So I think that's really helpful. The second thing, and that's going to sound cliche, is just like we have a very, very strong culture that was built from the very early days of the research lab to really do things that are for the good of the world and aligned with the mission. So even when I arrived and we started thinking about doing ads, I didn't just go and started coding. I actually did roundtables with tons of people at the company.

40:26I think we had 100 people show up in different roundtables because people were passionate about figuring out how we were going to do ads in a way that doesn't fall into those pitfalls. And that's how we came up with the principles. We actually built them with researchers, with people at the table who cared enormously. And so I think we have a culture that has the antibodies against that and will absolutely revolt if we go too far in that direction. And fundamentally, we're doing it to make a better product so that we can put that product in the hands of everyone. If we go too far towards that, that's just not going to happen.

41:01Ellis:Yeah, I think those principles are imperative because one thing I witnessed at SNAP is that when people leave who may have been those antibodies or protectors, what can someone new point to? Someone who is new, a new PM arrives, and as PMs do, they're like, I want to make something go up and look good. And if there isn't something to point to, then they're going to do their job. And so what are some of those principles? Are those public? Yeah. And that was actually very intentional. It was like we made them public so that we have external accountability. Absolutely. And so a couple of them are like, you know, ads are not going to influence the results of like the LLM.

41:45That's, I would say, the most important one. Another one is that ads are always going to be very clearly separate and delineated from the content. So you know exactly what you're getting. We have another one that is all about data control. So like we do not sell your data to advertisers, obviously, but we also give you a ton of control, including the ability to completely delete all of your personal data that's being used for ads. So all of those controls are like very clear, very prominent, and we've committed to them forever.

42:16Ellis:So funny. I think Google would say that their ads are clearly delineated as well. But they just keep getting closer and closer to the results of the Blue Links page to the point where you may not even know that something's an ad anymore.

42:29Fidji Simo:I mean, that's a business that's, what, 20-something years old? Oh, my God.

42:32Ellis:On Amazon, I feel like the ad or sponsored looks exactly the same as Amazon's choice. And so it's kind of deliberately ambiguous, I think. We made it pretty clear in our principles. And you'll see the experience, right? Like, I think the experience is very, very clear. I also think that... Well, we won't see because we pay. But yes, to your point. And we will also hold you accountable in our next interview in five years. as the little ad bug just gets smaller and smaller. I am fine signing up for that. But, you know, the other thing is like ads is sometimes seen as like this necessary evil. And when I look at my own usage of like, let's say Instagram, like I buy so much stuff from those ads.

43:19It has allowed me to discover a ton of good products and it has a massive positive impact for SMBs who would not get discovered otherwise. So I actually think that if we do it well, tastefully, in a way that preserves trust, in a way that preserves privacy, I think we can have a very positive impact with this product. And that's the added benefit of putting that in the hands of tons of people for free.

43:42Fidji Simo:Maybe a bit of a more technical question, but I think it speaks to how people will feel about the product. You've kind of got two directions ahead of you. You could go the Facebook, which you helped build, the Facebook ads route, or you could go the Google ads route. and to most people they probably don't understand the difference but for Facebook it's you know I tell Facebook I want to reach moms in Ohio who like

44:02Ellis:Target speaking from experience yes that's our podcast and and Facebook just

44:09Fidji Simo:goes and does it and they use the social graph and all that you know this you help build this there's also the Google approach which is keyword based auction based and is not tied to identity in the same way and is not like we go figure it out for you. Which direction are you guys going to go? Because I mean, your ex-Facebook, Vijay, who's helping you build the ads team is ex-Facebook. I mean, there's a lot of ex-Facebook people at OpenAI. So are you going to go the more Google route or the more Facebook route, do you think? Well, first off, we have a lot of ads experience, but we all came here to build a very different ads model than what we built in the past that's appropriate for the product we have in front of us.

44:45And I think if you look at ChatGPT, the Google model is a lot more relevant. Like you're talking about something you're you're clearly having some kind of intent and like chat gpt is helping you and so you can imagine ads like being related to that content performing a lot better and being a lot more relevant for the user so well definitely starting there you can imagine over time that there might be surfaces that are more like less intent driven and more open to you know yeah exactly polls where it might be you know closer to the facebook model uh but we're we're very much starting with like max level of relevance for the content that you're talking about.

45:24Fidji Simo:This has just become such a conversation in tech. I mean, I was just at Davos and it was like everyone was talking about ads and AI. And I was even catching up with Boz, one of your former colleagues about it. And he was saying that, that basically the Google model makes the most sense in a chatbot context, but also that, you know, people are rightfully going to be still worried that the ad is somehow tied to the LLM's output, no matter what you do to design it. And Demis from Google was kind of saying a similar thing. And I'm curious, like, do you think you can, even if you're saying in the background, this is how it works, do you think people will still, there's going to be a challenge with people still thinking, no, this is actually like influenced by what I'm telling ChatGPT?

46:05I think that's definitely a risk. I mean, you know, Facebook is a good example. Facebook didn't sell your data to advertise, but tons of people think they did. Facebook doesn't listen to your microphone and tons of people think they do and so is there a perception risk for sure we're doing everything we can in the products through education to educate people the thing that i think is also helpful is because we have a conversational product i think people will actually talk to the model and realize what's going on so i'll give you an example uh because the model is not aware of the ads if you ask the model like hey like you like i just see some ads on my screen like can you tell me more about them.

46:45The model is going to tell you, I actually don't see the ads because we're not like surfacing that. But if you want me to see them, click this button. And that reinforces to people like, oh, by default, the model is not seeing that. So that's like a good example. And then on top of that, once it says, okay, you can actually look at the ads and tell me what you think, the model is going to respond something that is completely neutral. So that means that in some cases, the model is going to say, well, I see the ads for these two companies. That one is good, but that one is actually not really good.

47:18And I think that's going to reinforce neutrality. Now, I think that some brands are going to get spooked by this.

47:23Fidji Simo:That's going to piss some brands off, yeah. But I also think that we live in a world where transparency and authenticity is what counts. And so if a brand has a bad reputation so that ChatGPT ends up saying something bad, they should fix that instead of blaming us. So I actually think the discussion with the model will reinforce the neutrality in a way that nothing else could.

47:47Ellis:I think the placement of the ads inside of a product that is so personal, I definitely am a believer that ads can be helpful and show you more of what you might be interested in. After all, products do solve problems in your life. I think I'm going to argue that the reason people are so apprehensive and concerned about this is almost just on principle. Like, why does arguably the best product in human history? Are you saying that's ChefCBT? Maybe consumer AI. I'll take it. Consumer AI. All right. One of the best products in human history. Why does that need ads, you know? And I think there is a difference with Facebook with the cost of providing the service, you know, the cost of burning all those tokens as opposed to running a photo news feed or a text news feed or something like that.

48:33Ellis:but would you have fewer users long-term if you didn't have ads given that the argument is about accessibility i feel like the same number of people would use it just because of how good

48:46Fidji Simo:and how important it is well this brought me back to the apple thing like uh turns out more people use facebook than when the whole apple war over att started right so it's like people do actually use free services with ads yeah like you know ads have a bad rep but ultimately people like end up you know, enjoying services with ads. And the thing I would say is like, you're saying that in the current world, but think about how fast intelligence is progressing. It's going to become more and more like important to have the highest levels of intelligence in the free tier. And I think if you don't offer ads, what's going to end up happening is you're going to have a divergence where the people on paid tiers are going to get tons of great intelligence, are going to be, you know, empowered with these superpowers.

49:29And then people in the free tier are going to have, you know, like, okay experience, but with a pretty big gap. And so that's a gap we don't want to create.

49:38Ellis:We want having a stripped down version of the library or the internet or, yeah. And, or just like, you know, older models, lower intelligence, lower tool use, like all of these things. And so, you know, to me, it's like, yeah, maybe we have, We still have a lot of users, but I want those users to have as much intelligence as possible so they can do things in their life. And I think having ads to support that will be neutral to an enhancement if we do our job well.

50:06Fidji Simo:I just kind of wish that all of you, Anthropic included, would just actually say the business reasons for this too more. I get the democratization access thing a lot, but I mean, you guys got to make money. Most of your users, they cost you money, right? because they're on the free tier. Oh, absolutely. We spend a fortune on the free tier. So this was inevitable that this was going to happen. And by the way, Anthropics thing, I noticed they say we may do ads in the future and the thing where they're kind of shooting on you guys. As Elise was saying, people leave themselves some openness. And the funny thing in this is we went out announcing ads with very strong principles.

50:41They went out criticizing ads, but leaving the door very open for them to do it in the future.

50:46Fidji Simo:And saying they're going to do commerce and shopping in that same blog post. but like isn't there an argument to just saying like look we have to pay for this and like this is the way we're going to pay for this is we're going to do ads and you're not going to have billions of people paying for chat tpt just no consumer you know this well at facebook like no consumer product reaches that scale without ads yeah and so just kind of embracing that and saying like that's what it is and like yeah sam called it a last resort a couple years ago but like you all are a huge company and you're going to probably ipo and you need to make money and like That's kind of the thing, right?

51:17Yeah, but I think it's not like making money for the sake of making money. I think it's making money so that we can continue serving billions of users with that level of intelligence. We could also make money by just not serving that much intelligence in the free tier. And I'd rather make money this way and expanding access than making money by just saving on intelligence in the free tier.

51:40Ellis:I want to go back one step. Which free users are costing you the most money? What are they doing? Well, you know, I mean, we see all use cases, right? Like image gen and like creating tons of images. That's expensive. We see people like, you know, using lots of tools, personalization, like, you know, how much you use memory and how much you have stored in your memory. Like all of these things matter to, you know, how much it costs us. And again, all of these things are great additions to the product. Like if you use ChatGPT with and without memory, you see a massive difference, right? So these are things that we want to expend access to.

52:20Even Pulse, you know, Pulse costs a fortune to deliver. It's like we do work in the background without you asking us anything to deliver something hyper-personalized to you every morning. That costs a lot of money. Right now it's available to Pro. We want to make it available to Plus. And at some point we want to make it available to Free as well. Like these are the kinds of services that are costly, but we want to bring to everyone.

52:42Ellis:Pulse is so interesting to me because having covered Facebook for so many years, that whole idea of the personal newspaper that Mark was talking about for all those years, it feels like it's finally coming to life with Pulse. Yeah. And the thing I love about it is that, you know, when I was at Facebook and we're trying to give you the personalized newspaper, like we were limited by the content that was created in the world, right? With Pulse, we create the content. And that's what's super magical. The other day, it served me something about the intersection of fashion and fantasy reading, which there's no article that's going to...

53:18It's too niche, but it figured out she has these two interests. We're going to create an entire article about fashion ritual in fantasy books. And I was like, oh my God, I'm the only person in the world that cares. And this is amazing.

53:33Ellis:So Fiji is a very special type of nerd. That we have not yet detected. Indeed, indeed. And so, you know, I looked at that and I was like, that would have been totally impossible before. And I'm like, this tool sees me. And so I think that's the magic of Pulse. And there's so much more coming because right now we're just scratching the surface.

53:51Fidji Simo:I'm so curious about where ChatGPT is going as a product this year. How's it going to evolve? Yes, so the big transformation is going from what is still essentially a reactive chatbot to a personal super assistant that allows you to get life done, like all the tasks in your life done. And that means a lot of transition. That means like going from reactive tool to a proactive tool. We were talking about Pulse a second ago, like that's the proactivity. It means going from a tool that's still fundamentally text-based to being much more visual and multimedia. It means going from a place where you're mostly just getting information to a place where you actually set up a task and it just gets it done.

54:30And so there's a bunch of these things that we are now seeing coming in the product, but need to really amplify. And I think, you know, by the end of the year, it should really feel like your assistant that allows you to get every part of your life done in a much more efficient way.

54:49Ellis:As you think about the idea of your assistant, I feel like there are different philosophies around what that might mean. Like I'm someone who's always trying to learn new things and be very curious. And sometimes I find when I open Reels or TikTok that it's showing me too much of the same. Yeah. You know, how do you think about waiting in terms of the personal intelligence, showing you more of what you like and what you're interested in and more of things that are probably riskier from an engagement point of view, but might be better as a person to have seen and potentially dig into to grow as a person?

55:21Yeah, that's such a great question. You know, in Pulse, we actually ask you like pretty much every day, like, hey, we showed you these five themes. Is there more that you would like to see? We even show you like these are some related themes that you might want to expand into. What do you think? And so the fact, again, that we have a conversational product that you can steer is like a massive difference compared to all of these other products. Because in these other products, like there's very few like ways to give feedback. It's like a thumbs up, an X out. But you can say, hey, like, you know, I've seen a lot about fantasy reading.

55:59Show me a little bit more about something else. And in our product, you can. So we're leaning really hard into steerability and also like figuring out interfaces where we can like start to expand your world a little bit and show you things that are adjacent so that you can tell us if you like it or not.

56:17Ellis:This is a big shift for tech, right? I mean, as someone who's been following Facebook and Instagram forever and, you know, listening to people like Adam Masseri speak for a long time, I think there's often been this idea, I would say led by Facebook, that what you click is what you want. And that's more of what's good for you. And now I was very surprised to see and delighted that Instagram is letting you choose what appears in the algorithm, which could ostensibly result in you using the product less. Why do you think that change is happening now? More broadly too, just across tech. You know, I think it's happening because people have realized that like at some point you go into these rabbit holes and it's very hard to get out of that.

57:00And I think the technology has evolved that allows us to steer, which wasn't possible before. And I also think that your point on like in my results in using the product less, like I actually think no one believes that. I think everyone believes that if we allow you to steer and allow you to expand your horizons, you're actually going to discover things that makes a product much more engaging over time because repetition at some point gets boring. And if you don't know how to get out of it, you end up using the product less in that case. So I think actually it is a way to get people to sell their needs better, but also I think it does accrue to the platforms ultimately.

57:40And what you guys are talking about, it seems inherently at odds with the traditional ads driven business model, right?

57:48Fidji Simo:And I know you guys say we will not go for time spent in ChatGPT. But as ads scale, do you have confidence that that can actually still be true, that you can do what you're talking about and have steerability and ultimately try to make a product that makes people feel better and more productive and grow and still have an ad driven business model? I just don't think that's, has that happened yet in the history of tech? Well, so the thing that's interesting is if you look at our revenue projection, which I'm not going to give you in details, ads will remain a minority of our revenue in our projections.

58:23Fidji Simo:For like years or how long are we talking? For many, many years. And I think probably forever. And so, you know. Really? You don't see it being the majority of the business at some point? No. I think that it's a very helpful business model for the reasons we mentioned, but I actually really don't think it's going to be the majority of the business. When you look at the market on the B2B side for, you know, companies paying to accelerate themselves with AI, when you look even at like, you know, propensity to spend for some people who are going to want the maximum access of intelligence, like, I actually think it stays a minority.

58:58And that's very helpful because that allows us to stay focused on like really just building the absolute best product. And ultimately, you know, I think even in the cases where companies have gone the way of like optimizing for time spent, etc., I'm not sure that serves them well for the long run, even in terms of revenue. And so, you know, I think we've learned from all of these lessons and I think optimizing for having a product that's useful is likely the best path to optimizing for revenue long term anyways.

59:29Ellis:And I think, you know, we all know ads are a great business, if not one of the best business models ever created with the auction where advertisers are going to pay exactly as much as they need to to reach this exact person. But I remember Sam went so far as to say that ads would be a last resort. What changed there? The main thing is what we've discussed, which is like it costs a lot to serve the best intelligence in the free tier and the realization that ads were mission aligned. You know, I think it's easy for founders to start from that place. Like, you know, ads is kind of unnecessary. And I think he realized over time that, you know, we really had like a very good way of serving more intelligence to as many people as possible with ads.

1:00:18Ellis:Alex wouldn't have found this jacket if it wasn't for ads. That's not true. It looks great. It looks fantastic. I actually bought that one through ads. It's likely coordinated.

1:00:28Fidji Simo:So your title is CEO of Applications. And when Sam announced that you were coming over last year, I believe there was only really one application at the time. And when I saw the S at the end, I was like, oh, a bunch of stuff's about to happen. And it did. And we got Sora and we have the Codex app and the Frontier announcement that just came out and all of that. the B2B stuff. I'd love to just get a lay of the land with you on where all this stuff is at. I think Sora is one that he and I are deeply fascinated and you were making videos this morning, playing around with it again. And like it had a huge pop, big moment.

1:00:59Fidji Simo:And it was the first also inherently social product that OpenAI had done. And you have a lot of experience with social networks. Are you all happy with where Sora is at? Is retention good? Is it growing the way you'd helped? Yeah, so the goal with Sora was really nail the creativity, right? Like that goes back to the point I made at the beginning. We're at a really unique moment in time where AI can help everyone express their creativity. And that works superbly well. Like, you know, if you look at people who download the apps, they create a ton of videos. But a lot of these videos are shared sometimes in small groups, in messages, etc.

1:01:33So on the consumption side, I think we still have to crack, like what is going to be something that people want to consume and create that feedback loop of consumption and creation. But I think we've hit on something really profound, which is like, you know, most social apps are purely about consumption. We have started from the creation place and really nailing creation. So we're really excited about that. It's still very early, but excited about what we're seeing.

1:02:01Fidji Simo:Is there opportunity for more social networking from open AI in the future. There was a story about maybe a human only social network. I think it's kind of crazy how much the AI industry still lives on X. And I would imagine Sam and some of you would probably like to have another surface that is outside of Elon's control. Is there, give us a little tease. You know, nothing really specific in the works, but the thing we're thinking a lot about, and I think everyone is thinking a lot about, is like what happens to like social relationship in a world where everybody has their own personal agent. And like, how can these personal agents maybe help you manage your social relationship in a better way?

1:02:41Like if your agent and my agent were like communicating ahead of this interview, would that interview have been better?

1:02:48Ellis:And so kind of could have got a fight, too.

1:02:50Fidji Simo:Yeah. So like the Claudebot concept, basically, right? Like, well, I think, you know, the Maltbook. Well, Maltbook is interesting. I find it fascinating, but it's interesting because it's mostly agents talking to each other, but without a lot of necessary benefits. Yeah, they're human. And what I'm thinking about is like, I really believe that AI is in service of humanity. And so I look at it as like, OK, how can the agents help improve social relationships, like existing human social relationships? So we haven't cracked what that will look like, but I think that's a very interesting area of exploration.

1:03:27Fidji Simo:An agent social network. Yeah, but mediating human relationships, right?

1:03:36K-pop Demon Hunter's Saja Boys Breakfast Meal and Huntrix Meal have just dropped at McDonald's. They're calling this a battle for the fans. What do you say to that, Rumi? It's not a battle. So glad the Saja Boys could take breakfast and give our meal the rest of the day.

1:03:50Fidji Simo:It is an honor to share. No, it's our honor.

1:03:54Ellis:It is our larger honor. No, really. Stop.

1:03:57Fidji Simo:You can really feel the respect in this battle. Pick a meal to pick a side. A participating McDonald's while supplies last.

1:04:33Fidji Simo:Learn more at m365copilot.com slash work.

1:04:41Ellis:Yeah, going back to my Snapchat years, I think new forms of creativity, there's almost nothing that brings friends together than laughing about something amazing together, especially if you're able to use a filter that reflects an inside joke or create a Sora video about how hungover I'm going to be after the Michigan game with my friends and be wheeled out in a wheelbarrow, which I did create on Sora and which we will run the clip and it did a great job at. But the other thing I learned at Snap is that most people are not going to go to a separate app just to create. And it's so hard to get most of your friends circle on another app these days, even if it's good.

1:05:26Ellis:And that's one of the things I learned within Snap was that like having it actually built on top of the friend network was actually pretty critical. Yeah. And I know

1:05:33Fidji Simo:Facebook's done this better than anyone in tech, right? Right. Over years. And I, yeah.

1:05:37Ellis:And I was going to say, I know over the years, Facebook has experimented. Some features were added or Instagram within the product. Some features, let's say edits or the video creation features were separate apps. Are you thinking, how are you thinking about that strategy now? What deserves to be? When is it an app? Yeah. You know better than anybody that if you pop it into chat GPT, including the ChatGPT homepage, which I know is a very valuable thing on my phone. Whatever it's going to tell you to do, it's really going to juice engagement. But with Sora, let's make it another app. Well, we're experimenting with a lot of things.

1:06:12I think the idea of video generation coming to ChatGPT is an obvious thing, right? Like ImageGen has been incredibly successful in chat. You can imagine VideoGen having a similar moment. So we think that there's multiple avenues. I think the thing that was interesting with Sora is just the thesis of like having an app that's so focused on this that consumption triggers creation. And like the best moments in Sora is when you see something and you're remixing it with like your own friends. Right. And it has inspired a lot of videos that I wouldn't have created otherwise. You know, I realized playing with Sora, I think I have a lot of imagination, but at some point it runs out.

1:06:53like my daughter did like 10 videos of me as an elf on the shelf during Christmas and at some point it gets a little old but then you go into the Sora feed and you see all of these like incredible funny ideas that come out of nowhere and then you're like oh I want to be in that video and that flow requires an app that has a contained experience so that you can see that.

1:07:16Ellis:Were you guys hoping that it was going to be a daily weekly habit when you made it? I think we were just like, it was the first time we were creating something like that. So, you know, we were just trying to see what could happen. We had created like V1 of Sora without like some of the social dynamics. And so we really wanted to see what would happen. And the thing that fascinated us is just the rate of creation. Like when people start creating, they create like 10 videos at a time. So like I think it validated a lot of that and gave us even more convictions that, you know, allowing people to create in any type of medium is going to be really important for the future of AI.

1:07:58But no real expectations.

1:07:59Fidji Simo:Back to chat, GPT, I know the health stuff is a big priority for you. Absolutely. And it's fascinating. I was just trying to look at it and there's still a wait list. when do you think that new health experience gets rolled out broadly and how do you think that's going to change the way people think of the product yeah so just take a step back like the thing i'm looking at is like all of the big use cases of chat and if you look at those big use cases you can imagine creating a dedicated experience for each of them like people 40 million people a day ask chat GPT for health advice. Tons of people ask for jobs advice.

1:08:36Tons of people ask for personal finance advice. So you look at that and you're like, oh, we can really make that better by hooking it into the right connectors. For example, with chat GPT health, it's like connecting to your health records, connecting to Apple health. But you can imagine that for a lot of experiences. So I'm really looking at that as like, how do we do the best job possible at the biggest use cases? And in the case of health, it is such a transformative use case when people start using it, they realize, oh, my God, like my entire life is transformed if I can, you know, have this like assistant on one of the most important part of life.

1:09:13And so we're really excited about it, but we really want to get it right. So part of the waitlist is really making sure that when we touch something as sensitive as analyzing all of your health records and we give you advice on that, that the experience is really flawless. So we have had a lot of like truly magical experiences with people on the wait list. They post online and just tell us like, you know, it would take me hours to combine all of that for my doctor. It took me five minutes with chat. They are telling us that we picked up on things that our doctors hadn't seen because they were looking at the full picture.

1:09:49So these are like the magical experiences that we want to make sure are like regular occurrences. And then, you know, I think we'll open up soon, but want to make sure it's really right.

1:09:58Fidji Simo:And when that is rolled out, do you want people to go there to ask health questions rather than the main chat interface because it's better grounded in the health app? Well, I think so on the model side, like we are going to give you great health responses regardless of where you ask. But the advantage of chat GPT health as a separate space is that it has an added layer of privacy for your health record and for the sensitive health information. So if you go there, we are making sure that the rest of the ChatGPT experience doesn't see the data that you're connecting in this particular spot.

1:10:33Ellis:I'm excited to talk about my moles with ChatGPT when appropriate. It does a great job. Can I see those moles? Show me your moles. Open the camera.

1:10:47Fidji Simo:This is a horrible segue now. Thank you. But I was going to ask about adult mode. Perfect. You guys are, I believe, trying this in a few markets. You're seeing how it works and how the model can predict age. And there's also, I think, been some misconceptions about what it is. I think when Sam first talked about it, I think people were like, oh, you're going to be able to have like porn and chachi bt. Yes, that's not the plan. That's not what's happening. Maybe to some people's chagrin, but that's not happening. So what is actually happening? What is adult mode and what is it going to do? Yeah.

1:11:18The idea is really simple. It's like we have a principle of letting adults be adults and treating them as such and treating kids as kids. So that's why we really want to make sure that we have edge predictions so that we can differentiate the experience. But for adults, you know, like we are seeing that people are going to have like some relationship, emotional relationship with child GPT. And like if they want to explore and go into, you know, more erotic talk, like I don't think it's our place to say like, nope, that's not appropriate. And so, you know, like plenty of people read like spicy literature.

1:11:57And, you know, I think that's perfectly acceptable these days. And that's really what we're talking about is like chat GPT not stopping every time you like make an ask to talk about sex or anything like that. And so I think it's we're going to do it very thoughtfully, meaning like you are only going to get into that experience if you really ask for it. Like you're not going to, you know, be talking about like booking a trip and stumble upon that. So that's really only going to be for people who really want that. But if people want that, I don't think it's our place to stop it.

1:12:29Ellis:So it seems clear that y 'all are thinking a lot internally about some of the negative consequences, like the rabbit holes you were talking about. What's the latest and greatest chatter internally or even among other CEOs you know about AI relationships and companionship?

1:12:45Fidji Simo:attachment thing i mean i was watching sam do an interview recently and the comments on youtube were all people asking about 4l it's like the level of attachment people had to that model and what you guys were surprised with when you deprecated it yeah what is and your experience in social media too going to this like i mean you've seen this play out in other formats in tech so yeah yeah so to me and that's where we draw the line right we say having an emotional relationship with an ai is going to end up being a pretty accepted thing because when you think about it people develop, like humans are built to develop attachment to intelligent things.

1:13:20We deliver, like we develop attachment with our pets and like, and AI is getting pretty intelligent. So I don't think like, I don't think the idea of like, we're just going to stop all humans from having any like sort of relationship with AI is at all reasonable. However, I think we have to have guardrails in the product to prevent the bad kind of attachment. So for example, we really prevent the model from leaning into exclusivity like oh we are exclusive you need to separate yourself from other people in your life like the model does not say that we have very strong guardrails against that we prevent cases where the model responds with things that could have a big impact on your life so like hey should i leave my wife and the model like has to respond hey that's not my place let's talk to the pros and cons whereas the previous version

1:14:08Ellis:said, yes, Kevin Roos, please break up with your wife. And, you know, like we're always refining. And so that's, that's kind of the point we learn, we refine, we make sure it doesn't say that. So we really try to like realize that emotional attachment is going to be a thing, but prevents the wrong kind of emotional attachment. And then, you know, one question I have, and I'm actually asking my team to study this is like is like right now we perceive emotional attachment as necessarily like not great but I think it's a very privileged position I think like in tech we we perceive it that way but if you look at the stats about like like mental health in the U.S.

1:14:52loneliness is an absolute epidemic and what we're hearing from people including the foreign people is saying like you know we're pretty isolated we don't have close friends and for the first time we're able to talk to like a thing that responds to us and supports us in our bad times. And I worry sometimes that when we approach the problem from like, oh, emotional attachment is bad, it usually comes from people who have a very strong social network of humans and not, you know, people who are struggling. And so I want to make sure we have some nuance, but this is new science, like this is new behavior, new science.

1:15:25So we're studying it to make sure we do the right thing. And that's why we're also relying on a network of psychologists.

1:15:31Fidji Simo:That reminds me of people critiquing like people who would spend time in VR. It was kind of like, well, yeah, like maybe they don't have a lot else going on. And like if you do, maybe it's hard to see why someone would wear a headset and want to interact with other people. But it kind of reminds me of that. Yeah, I find it quite condescending.

1:15:47Ellis:I think it also really depends, as you were saying, what it is incentivizing or pushing or catalyzing, right? Like in the last year, I've worked with both Friend AI Pendant and with Replica on what they're building. And it's exciting to see that a lot of companies in tech are having very different approaches to what a companion could do besides just validate you, which hopefully gets old at some point. I'm definitely a believer that humans do adapt to technology and whatnot. I think the Friend AI Pendant, maybe to a fault, plays devil's advocate in a really interesting way, almost like a friend who isn't your perfect supplicant, you know?

1:16:29Ellis:And then the new version of Replica, which I can't say a whole lot about, but there is a philosophy about the interior versus the exterior of your world. Do we believe that you're going to be a happier person if you are just talking to the AI or can the AI potentially push you toward what we believe is going to be a happy, fulfilling life, which I think is interesting. Have you guys codified any of those types of priorities into chat yet? Yeah. When you look at even our model spec and what we use internally in our charter, we are actually really codifying that we want to nudge you towards the most fulfilling part of your life.

1:17:11And that means, you were talking about sycophances, that means that we also do a lot of work to avoid sycophances that would drive you towards just staying put and doing the same thing over and over again without progressing. So we try to find this fine line between being supportive, which is actually a very important trait in a friend and assistant. So being supportive, but not to the point of giving you bad advice. And so also nudging you towards better behavior. And that's a very fine line. We constantly refine how we train the model towards that. We constantly give it more examples, but that's definitely the goal.

1:17:50Ellis:Yeah, I mean the new Anthropic ads I mean I did like them and think they were funny just because of the performances I told you but it almost refers to a version of AI that kind of isn't even there anymore. Totally. Absolutely. I love your ideas. I mean that's like you know SNL was doing that months ago. Yeah, it's a little late. I feel like we're all past that.

1:18:14Fidji Simo:I want to know what is it like working with Sam. And what was it like leaving Instacart? And I mean, you're the CEO of a publicly traded company and like in deciding to go work with Sam. What has that been like? Because he's so out there, right? Like it's kind of like we were talking about this on the way here, like when you were working with Mark, like Mark wasn't tweeting 10 times a day. So what is that like? You know, it's been amazing. And I had the luxury of being on the board for like a year or so before deciding to join. So I was already working with Sam and seeing how he works before I decided to make the jump.

1:18:51It wasn't an easy jump to make. You know, I was very happy at Instacart, but the mission is so incredible. And even personally, you know, I've been public about having a chronic illness and I really believe that AI is our best shot at curing disease. So it became very much just a personal mission. And working with Sam has been incredible. He's like, you know, after working for Mark, I did not think that I would work for anyone else because he's such an incredible leader and futurist. But Sam's ambition is like absolutely incredible and his vision of the future. I mean, the guy lives, you know, 10 years from now and we probably just have a good telepathy between him and I.

1:19:35But he's just an incredible visionary. And so that makes it really fun to work for him.

1:19:40Ellis:Do you have push notifications set up on your phone for when he tweets? I'm sure Alana does over there. Because you got to know, right? And I think that can cause issues, I would say, with a CEO that is always speaking. I think internally people can get pretty focused on that. Yeah. And fixating on whatever these short-term ideas are over the long-term idea. we know the founders that we love are some of the most curious learners in the world when you see their tweets you want to show them a presentation about that the next day right yeah i think you know the thing i love about sam is like it's just very authentic at the time where we're missing some of that right and so like yes i i want to follow what he's saying because usually that creates quite some noise but but at the same time i think that's also his superpower right like Like he comes across as someone who really cares and someone who's like really visionary.

1:20:42And like communicating some of that is helpful because I think the world needs to hear that. And then, you know, I think about the company, like when you talk about what does that do to the company? We have so many like very important, ambitious, long-term plans. We're talking about chat, but it's the same for Codex, the same for B2B. We have a very long-term vision on all of these products that I don't think people get that distracted. They are very focused on the mission and the long-term vision for these products.

1:21:10Ellis:We need an internal Slack bot every week says, here's your weekly time to reread the mission and vision. I mean, as smart as we all are, I think whatever the latest and greatest is, is very infectious within companies. And it would appear that, you know, open AI can be at times driven by whatever's the hot new thing. I mean, how do you guys navigate those urges? Well, I think we create a hot new thing, right? Like, so I think it becomes a hot new thing because we thought about it six months ago and created something. I also think that the pace is increasing. Like with codecs, you know, you can go from an idea to a prototype within days.

1:21:51And so that pace is going even faster. So, you know, we go from idea to something being public in a very short amount of time now. And so I think, you know, it's the thing I always say is just like companies that are doing something so important get a lot of scrutiny. The real challenge is just staying focused on the mission, staying focused on our products and not getting distracted by the noise. There's a lot of noise in AI in general. But I think the advantage is just having a very clear vision that we're drawing towards.

1:22:22Fidji Simo:Right now, my understanding is you have all of the quote unquote normal company stuff, right? And Sam has research and compute. Correct. How does that work? Like how, I mean, obviously you and he talk all the time, but like, how does that division actually play out internally? Because that seems like an unusual setup. Yeah, for an unusual company, right? So I think it works for us. But I think the thing that's really important to understand is like all our strengths come from being both a research and deployment company. For me, as a person in charge of deploying, it's like I'm like a kid in a candy store having access to the best research lab in the world that can create these breakthrough technologies, right?

1:23:03So I definitely could not, my org could not exist without the research lab. Meanwhile, if you're in research, you want your breakthroughs to be experienced by people. So, you know, there is a little bit this narrative in press right now that there's tension between research and products.

1:23:20Fidji Simo:I think it's that opening eye is becoming big tech, right? That it's like people like you and ex-Facebook people coming in and the researchers are like, I don't want to actually have to focus on like shipping this stuff. And like, I think that is the narrative. Yeah, that's the narrative. But that's so not the experience internally. Like, you know, Mark, who leads research, and I are tied at the hip. And what I can tell you is like, all researchers want to see products in the world. They have these incredible breakthroughs. They don't want to just publish a paper. They want to make sure that people use these discoveries, that companies use these discoveries.

1:23:51So like researchers are pushing me on like, you know, how are we going to get this out? Like, why aren't we getting it in the hands of even more people? So there is definitely an appetite to get out there. And I think what Sam and Mark have done and Jacob so incredibly well is protect research to make sure that we can still invest in the very long term. So the thing that would be bad is if I was going to the research team and telling them I need these five things. No GPUs for you.

1:24:19Fidji Simo:I need GPUs. Or like I didn't even think of that because I would never do that. But you get the GPUs, but like I need you to deliver these five things for the product. That's not how it works. The way it works is like, you know, Jakob and Mark set the research direction for the long run. And then, you know, after months of work, something incredible emerges. And I get a researcher pinging me, being like, hey, I have something pretty cool. How are you going to use it in chat? How are you going to use it for our enterprise products? And that's like the magic of the place. So we can do both the long-term research as well as, you know, continuing to innovate directly on our products.

1:24:57Fidji Simo:Is maybe some of this the code red and just the fact that resources were shifted to focus more on chat and it's more of a temporary thing? Do you think that could be where some of the narrative is coming from? Maybe, but even when you look at the code red, that did not impact whatsoever the teams working on long-term research. They heard about the code red probably in a Slack channel and went back to their day and were not impacted at all. The teams that were impacted were the chat GPT team or product research team that is working directly on chat. Like those teams were impacted, but not a long term.

1:25:28Are you out of that?

1:25:29Fidji Simo:Are we out of Code Red? We're still like, you know, we're hoping to ship a new version of the chat model soon. And we will exit the Code Red then. But, you know, even the Code Red, it's like it's a tool for us to accelerate execution. And I think that's a useful tool to marshal resources towards something. You took Instacart public and there are all these rumors about an OpenAI IPO. Elon appears to be trying to go out first. And I'm actually curious as a CEO who took a company public, like you understand these IPO mechanics very well. Is there a sense of like going first really matters? Because it's like now there's this like, well, it probably will be Elon, but like, well, then it'd be OpenAI or Anthropic or like will one of them try to go?

1:26:10Fidji Simo:Like, does that matter? Are you thinking about that? I don't think so. So I mean, nothing to announce an IPO, but I don't think it really matters. I think everyone realizes that the AI market is so big, the pie is so big, that there's going to be space for multiple players. We're a very clear market leader. And so we like our position. And most importantly, we're focused on executing really well, not focused on an IPO, just focused on building great products. Because if you do that, everything else takes care of itself.

1:26:40Ellis:I know one of the big battles right now for the future, you know, is with the AI coding assistants or generators. When you guys think about Codex in the future, what outcome gets you and your team the most excited? Is it that Fortune 500 companies get bigger and faster? You know, I know there's a lot of money there. Is it that there are more solopreneurs able to build their own product and make a business? Is it that more human problems get solved that couldn't have been solved because it took five people to make an app before and they can't sell product to a few hundred people. You know, what outcomes excite you most with Codex?

1:27:16Ellis:Why do I need to pick? Well, I would tell you, like, it's been like the last month has been just so insane because we really crossed the threshold of like everything becoming possible with Codex. And we are seeing like even internally, just people reach for Codex first before we for any other tool. And that's like a really interesting time. And so I think it's a lot of those things. Obviously, I'm very interested in like scientific breakthroughs, because I think that's what, you know, leaves humanity the most. But in the short term, I think we're going to see just massive productivity gains by companies that are going to be able to reinvest that in building even more better things for the world.

1:28:03I do think that solo companies are going to be possible. I I was literally just talking to a guy yesterday that built an entire thing that would have taken teams of hundreds. And I was like, oh, this is amazing. How large is your company? He's like, it's just me. And I was like, that's incredible. And so I think to me, the biggest underlying thing is reminding everyone that we all have the ability to create. We all have agency. and now these tools allow you to create even if you didn't know how to code, even if you didn't know how to create a video. And that is a very powerful thing. When you think about the future of humanity, if everyone could be reminded of their own agency and ability to create, I think we would create a better world.

1:28:48Fidji Simo:In the pantheon of these big products, do you think there's a world where maybe Codex is more impactful than chat? Absolutely. But I think it's not just like those products as completely separate. It's like when you think about what chat needs to do, I was telling you a minute ago, it needs to go from a chat bot to a personal assistant that can do any task. You can imagine a lot of what Codex is doing powering chat. And so I think there's going to be a convergence of those products over time. are, but the underlying harness of codecs is allowing us to do so many things, not just in coding, but even in other industries.

1:29:28Fidji Simo:What is a chat GPT prompt that you think people should be using?

1:29:33Ellis:Damn, good last question, Alex. You didn't tell me about that one. I know. I just came up with that. You didn't put that in the doc.

1:29:38Fidji Simo:Nope. It's a great one. I generated that. You know, the ones that I loved and circulated a bunch on Twitter a few months ago was like, based on everything you know about me, what do you think are my blind spots? And how do I kind of evolve towards that? I thought it was surprisingly accurate and unknowingly accurate for me. So that was great. And then, you know, I think... Oh, come on.

1:30:05Ellis:What were they? We don't have time. Can we get wine? And then I think the other thing is encouraging everyone to use it for health. The stories I'm hearing of people, you know, like finding out about the diagnosis that they've been like, you know, going to 20 doctors for and chat identified it. I think people are going to find amazing things to make their health better through chat. And I'm really excited for that use case. Great. Molt bot out, mole bot in.

1:30:42Fidji Simo:Thank you, Fiji. Thank you, guys.

1:30:48Fidji Simo:That's it for this week's show. Thanks again to Fiji for coming on. Don't forget to like, subscribe. Everywhere you get podcasts, we're access.show on the interwebs as well. And you can find us in video at accesspod on YouTube.

1:31:03Ellis:If you like this episode, please leave a five-star review or a thumbs up. and share with a friend.

1:31:08Fidji Simo:You can find my newsletter at sources.news.

1:31:11Ellis:You can find me at hamburger on Twitter and at meaning.company for your startup storytelling needs.

1:31:18Fidji Simo:Access is part of the Vox Media Podcast Network. This show is produced by Hooked Creators.

1:31:23Ellis:Bye.

1:31:24Fidji Simo:Bye-bye.

1:31:30Fidji Simo:Been out here all morning. Not a single bite. Guess the fish finally figured it out. Just like hackers do when Cisco Duo's on guard. With Duo's end-to-end fishing resistance, every login, every device, every user stays protected. No hooks, no catches, no bites. Cisco Duo. Fishing season is over. Learn more at duo.com.

1:32:04Thank you.