The Deep View: Conversations

In this episode of The Deep View Conversations, senior reporter Nat Rubio-Licht sits down with Ian Silber, Head of Product Design at OpenAI, to explore how ChatGPT is evolving for the future. 

Silber explains why designing for AI requires a different mindset than designing traditional apps. Instead of treating the model as something behind the interface, he says designers now have to think of the model itself as part of the material they work with. That shift changes everything from product decisions and user experience to ethics, safety, and human judgment. 

The conversation also covers Silber’s experience at Instagram, how that shaped his approach to building fast-growing consumer products, and how OpenAI’s design team is thinking about the next phase of ChatGPT, including more proactive and agentic experiences. Silber also shares how he personally uses AI in his work, from brainstorming design principles to prototyping ideas with Codex. 

Topics covered:
+ How OpenAI approaches product design for ChatGPT
+ Why AI changes the traditional design process
+ What designers can learn from fast-growing consumer products like Instagram
+ How ethics, safety, and responsibility show up in AI design
+ Why human judgment will become more important as AI tools improve
+ How Codex and image generation are changing prototyping workflows
+ What young designers should know as they enter the AI era
+ Why curiosity and adaptability matter in a fast-changing industry 

If you’re a designer, product leader, builder, technologist, or anyone trying to understand how AI is changing creative work, product development, and human decision-making, you don’t want to miss this episode.

Subscribe to Deep View Conversations for interviews with the leaders shaping the future of AI, business, and technology. 

And don't forget to sign up for The Deep View daily newsletter. We don’t just cover AI, we decode it. In a world flooded with hype, we deliver sharp, no-nonsense insights to keep you ahead of the curve and help you put AI to work every day: subscribe.thedeepview.com 

Creators and Guests

Host
Nat Rubio-Licht
Senior Reporter at The Deep View

What is The Deep View: Conversations?

From frontier labs and enterprise platforms to emerging startups reshaping entire industries, The Deep View: Conversations podcast interviews the brightest minds and the most influential leaders in AI.

Ian Silber: I think you have to think about the model itself being a material that you work in. So instead of really thinking just about the pixels or the UI, we do think a lot about, like, okay, well, at the end of the day, what people really care about is the model or the output or the way you work with it. And so we think a lot about how we can make that experience better, right? Partnering, of course, with Research, which focuses on the behavior and the model, the personality and that sort of thing. And so partnering with them and thinking really deeply about that whole experience, because it's not just about, like, you know, where should this button go, how should this work, but really, what's the whole system and how does this all...

Nat Rubio-Licht: Thank you so much for taking the time to join me today, Ian. I really appreciate it. Let's just start by talking a little bit about your job. Tell me a little bit about OpenAI's design approach.

Ian Silber: My job is I oversee the product design team here at OpenAI, primarily focused on ChatGPT. Yeah, design at OpenAI, I mean, it's changing and evolving a lot in some ways. And then other ways, it's very similar to, you know, traditional product design. But, you know, I think at a high level, what we do is we try to understand how the technology is evolving, understand how our users want to use it or could use it, and try to package that up in a way that is approachable, that makes sense, that serves the many different types of users that we have, whether it's someone using this on the go, on their phone, to a professional user that's using this at work, using it to code or build something, and thinking through all of the different ways you could possibly use this, and trying to put that together into a cohesive product.

Ian Silber: Yeah, and in general, it's a really fun place to be because there's always some new thing, there's always some new tool or research breakthrough, and we get to figure out how to give that to people. So it's a little bit like being a kid in a candy shop.

Nat Rubio-Licht: How do you personally leverage AI in your job?

Ian Silber: So, I mean, honestly, it's changing every day. I kind of find different ways to do it. My job these days is less hands-on design and more helping to connect the dots and bring people together, and build the team, hire the team, help set the overall direction. And so at every step of the way, there's opportunity, right?

Ian Silber: You know, right now specifically, I'm working on thinking about principles for how we design. As your team grows, you need to make sure you have some cohesive way of approaching problems. And so I've been using it as a thought partner to just figure out what are our principles. I'll talk to ChatGPT a bunch and share the problems we're working through, what has worked, what hasn't worked, and have it brainstorm with me. That's a concrete example of some of the things I just used it for, to figure out how to concisely communicate what our principles are.

Ian Silber: And then, again, every step of the product development process, we find really fun ways. The other day, too, we were stuck on a design problem. And I really don't get a lot of opportunity to design anymore because I'm running around doing many other things, and I hire great people to do that. But I'll turn to Codex sometimes if I have an idea, and it can now prototype that for me. Something that would have taken me hours, I can now do really quickly. I'll send that as inspiration. So I still get a little chance to do some kind of quote-unquote design. That was a fun thing I did recently, and then I shipped it into the product.

Ian Silber: And so using it for that kind of thing, using it in the hiring process, we use it a lot to help with writing and synthesizing feedback. Definitely in the building process, too. There are so many different places at every step of the way.

Nat Rubio-Licht: Yeah. I know that you don't get too many opportunities to design anymore, but I want to talk a little bit about your background. You were at Instagram for a long time. How does that experience translate to building for OpenAI?

Ian Silber: Yeah, it's a good question. I think at Instagram, I learned so much from just this amazing opportunity to join when it was this beloved product that was growing. One thing I really learned was this attention to detail and craft and simplicity. And not just from a pure pixel or design level, but really understanding what it is we're building towards and having a really principled approach to how we build and really understanding who we're building for. That's something I try to carry forward, having those principles and thinking about exactly how we want people to use this and having some point of view. Kevin and Mikey, the co-founders, I think just had this amazing care for their product, and so that's something I took away as a designer.

Ian Silber: And then, yeah, I think just the opportunity to work on something that's growing so fast, there's competition, the landscape was changing a lot, right? In some ways that feels similar to now. We had Instagram, and then Snapchat comes out, and how do we respond to that? How do we stay true to what Instagram is, but evolve it? That was a fun challenge. And then similarly with TikTok. Having gone through a few of those, I think helped me understand a little bit how you can grow a consumer product that went from, I think when I joined, maybe a hundred million users or something like that, and now there are billions of users. Being able to see that ramp, and also seeing how the team grows, I learned a ton from that set of things too.

Nat Rubio-Licht: Yeah, your second time being at a major product with significant competition.

Ian Silber: Exactly. Yeah.

Nat Rubio-Licht: So how does designing for AI differ from designing for traditional apps and websites and things like that, since you've designed for traditional web apps and you've designed for AI?

Ian Silber: Yeah, it's super interesting. In some ways, Instagram had this property where we were designing the tools and the distribution or the shell for the content, but the content was all user generated. So we would try to get out of the way and make the simplest and best experience for you to see the content or to create it, and give you really simple tools there.

Ian Silber: In some ways, I can draw an interesting parallel to how we think about ChatGPT, which is: we want to connect you to all of the power and intelligence of the model and do that in a way that is most advantageous, easiest to use, and gives you the most intelligence out of it, helping you get there faster.

Ian Silber: The difference with AI, though, is a couple things. One, it's non-deterministic, which is in some ways similar to user-generated content, but very different in that you really don't know, you can't really control exactly what the output is going to be. That makes for a really fun, different way of thinking as a designer and as a builder. I think you have to think about the model itself being a material that you work in. So instead of really thinking just about the pixels or the UI, we do think a lot about, okay, well, at the end of the day, what people really care about is the model, or the output, or the way you work with it. And so we think a lot about how we can make that experience better, partnering closely with Research. We have a really cool team here called Model Design, which focuses on the behavior and the way the model, the personality, and that sort of thing. So partnering with them and thinking really deeply about that whole experience, because it's not just about where should this button go or how should this work, but really how do, what's the whole system, and how does it all work.

Nat Rubio-Licht: Now it's time for a word from this week's sponsor, Microsoft. What does it take to go from an AI idea to your first paying customer? Most teams are experimenting with AI, but experimentation doesn't generate revenue. Shipping does. Microsoft AI Envisioning Day is a free video series built for developers and software companies ready to turn ideas into real products. You'll learn how to take an AI idea to a working MVP that customers will pay for using practical patterns and guidance so that you can build something that actually generates revenue. No fluff, just clear frameworks and steps you can start using right away. If you're building with AI and want to start closing deals, not just shipping demos, this is how you get there. You can get started today at aka.ms/thedeepview. That's aka.ms/thedeepview. And we thank Microsoft for their support of The Deep View. And now back to the show.

Nat Rubio-Licht: Yeah, that leads me to my next question. Behavior is something that is important, and a big part of my beat is thinking about AI from the perspective of safety and responsibility ethics. How do ethics and responsibility come up in the design process for ChatGPT and OpenAI?

Ian Silber: Yeah, I mean, we take it very seriously as a key part of how we think about developing these products. At the core of our mission, we want AGI to benefit all of humanity. And so we're always thinking about, as we're releasing new models, making sure we're going through a very rigorous understanding of how people might use them. When it comes down to the actual design, of course, we want to think really deeply about both ways people could use it, and making sure that we're taking a very balanced approach and thinking very responsibly about it. It's important. We want to give this to as many people, but we want to make sure the benefit is there. So I think always tying it back to that mission and thinking through that. Like I said, we have very rigorous processes for that.

Nat Rubio-Licht: So like you mentioned, Instagram gave you this background of building a really fantastic product for consumers. And it's obviously part of the mission of what you're doing with ChatGPT. I guess, how can we boost adoption, or how can we get people that are not in the world of tech to adopt AI? Because obviously people like you and me are maybe a little bit more in the bubble of this stuff. So how do you get out of that? How do you get people to adopt it?

Ian Silber: Yeah, totally. It's very easy, you know, you go on Twitter or X and you see, "Here's my 10 prompts to make me a better whatever." And we don't want that to be the way that people discover how you can do this thing. I think that's a big thing that, honestly, we haven't fully cracked in the product itself.

Ian Silber: One thing we think about a lot is we're on a really big shift. I think that the last couple of years, the models have gotten better and better and better. They're really good at things, and they're getting better and better at things like writing and information seeking and research and thought partnership. But now you see with things like Codex, this really feels like a step change. We're thinking deeply about how we can help everybody understand how to get the value out of that. That's a really fun design challenge.

Ian Silber: I think now that these models can spend a lot of time and really work hard to get an actual task done, not just answer your question, but really go and do some work for you, the fact that they can run on this computer and go off and do these meaningful tasks, they can write code for you on your behalf without you ever having to know that's what it's doing. They're getting really, really good at doing things. So now, yeah, our opportunity is to demonstrate and show you how you can actually do that.

Ian Silber: I think we're working on a lot of great things that will come to ChatGPT that will make that much more obvious and evident. Thinking about chat shifting from being something that you have to come and ask for always, to something that can be more proactive, is something that we're really excited about.

Nat Rubio-Licht: So I feel like this conversation sort of lends itself to this idea that we are walking in this direction of AI becoming a background technology, which is something that I've heard a lot of conversations about. This idea that eventually it'll turn into something like electricity or the internet, where it's the foundation for a lot of different technologies and something that, like agents, maybe we don't even think about one day. What's your take on that? And do you think that that's something that is possible, something that's safe?

Ian Silber: Yeah. I think that is definitely a clear evolution where people will be able to leverage AI working on their behalf to help them achieve their goals or do whatever it is that they are set out to do. And yeah, this idea of this ambient AI or intelligence seems really cool. I'm really excited about the idea that we can give that power to people that otherwise wouldn't have that. As the technology becomes more capable, but also comes down in cost and we can actually deliver that, I do think it will give people this incredible tool. Again, whatever that is, the idea of it being able to understand your goals and really help you achieve those. Going back to: right now, I have to come to ChatGPT every time I want to do something. If it could work on my behalf, that would be amazing.

Nat Rubio-Licht: Yeah. So I have got one more question for you, and it's a question that I like to ask everyone that has their head in this industry. What is your AI hot take?

Ian Silber: I mean, I have many. I guess I'll talk about one that's close to me because it's about my role. I think everyone's asking, what does this mean to be a designer, or an engineer, or a PM, or whatever? There's lots of discussion around what's happening to all these functions, and can we just have one person do everything? I do think that, 100%, that is totally the case. But I also believe that there's a role for people that think differently about different things, and using these tools with your unique perspective or skill or background.

Ian Silber: In some ways, I see the design role changing a ton in that you have way more tools. But I also see that, in some ways, in a design role there's still a unique opportunity for you to really approach this with the mindset of a designer, which is: what is the system? What is the problem? Who are we solving it for? The humanity side of who are we building for, what's our unique perspective on it, I really don't think that's going away. I don't know if that's a hot take or not. I don't think that every designer needs to become an engineer and every engineer needs to become a designer, but I do think that the tools are helping you be whoever you want to be, which is kind of cool.

Ian Silber: In my career, I've seen many different people of many different shapes, things that they spike out that they're good at. I think this hopefully just helps you find more of that, go deeper into the things that you care about. I still think there's going to be a really, I don't know, I just think there's more and more opportunity now for people because they have these tools to go and build or contribute in the way that best suits their strengths.

Nat Rubio-Licht: I'd love to talk a little bit about how the role of the designer is changing. Like you mentioned in your AI hot take, this idea that anyone can do anything. Because I come to things from the point of responsibility and safety, are there potentially issues with the fact that anyone can do anything?

Ian Silber: Yeah, I mean, I think with all technology, it's a double-edged sword. For sure, you have to think about both sides of that. Of course, we take all that very seriously. But I think that roles are evolving quite a bit. For example, a designer on our team used our new image gen model to help riff on some ideas for a bit of UI, and then they're able to put that into Codex, and then Codex was able to code it up and submit it to the code base, which is just a completely new way of working that you couldn't have done a month ago.

Ian Silber: One thing I think that's really important for any role is adapting to how quickly all of this is changing. If I had any advice, it would just be curiosity. Playing with these things, trying them out, seeing what the tools can do, where they break down. We're finding all these new ways of doing things. So I think for any role, trying these things out and being adaptive to how things are shifting is probably pretty beneficial.

Ian Silber: As a designer, one thing I think is interesting to think about is that because designers now theoretically could just be writing the code, or not writing, but using Codex to do the engineering part in some ways, you have to think about, well, what does that mean that they're not doing? And thinking about how you find the right tool for the right part of the process. A lot of the design thinking happens on a whiteboard or sketching, and it's a very messy process where you're trying things out and throwing them away. So for our design team, we try to use whatever tool helps you get to solving the problem best. I would advise that for any function, whether you're a product manager or an engineer or sales or whatever you do, this stuff is changing day by day. We just try to be really adaptive and know that even something that might not work today might work tomorrow, and our workflows are constantly evolving.

Nat Rubio-Licht: Yeah. So I'm wondering, are we getting to a point where young designers, young coders, young people in any job function are going to need to become jacks of all trades as far as leveraging AI?

Ian Silber: I think, yeah, that's a really good question. It also depends a little bit on where you're working and the overall team structure. At a bigger company, maybe you can afford to have more specialists. That is a benefit of having a team of scale. As I think about growing our team, we can hire someone who's an incredible visual designer or incredible systems thinker or whatever. At a small startup, you don't really have that luxury. That's one benefit of these tools. It can help everybody be a jack of all trades. It can also help someone who's an incredible visual designer here round out their abilities. So yeah, I think that it can hopefully augment you in whatever shape you are.

Nat Rubio-Licht: I try to be as thoughtful as possible about the ways that I use AI, because I'm a writer and I know that AI can write. I intentionally don't use it for that, because that is something that I love doing with my brain and my hands. I'll use it, however, for the things that I am not so good at, that I know I need help with. For example, headlines. I am just a bad headline writer, my editor would tell you that as well. And I'll use it for helping with picking through research. For example, I wrote a story this week about how OpenAI just released this new networking protocol, and it was a very dense paper. So I used an AI model to help me weave through that. And then I went back and checked all of its work, because being a journalist means you go and double check everything. And sometimes I'll use it to help me surface links, like, give me 10 links about this news topic. I find that it does a really good job of surfacing links as well. So I have some very thoughtful, specific use cases for it. But that actually leads me to another question: are there places in the design process where human judgment and keeping a human in the loop is particularly important?

Ian Silber: I think so. Yeah. At the end of the day, everything that we do is a human decision. I don't see that changing anytime soon. Especially as more and more people can build more and more things, a lot of our job will be understanding how all the pieces fit together, saying no to a lot of things, and figuring out the thing that is the most important or the best way to do something. All day long, it's all human interaction. AI helps us build this thing, but at the end of the day we are there thinking deeply about how this is going to actually feel when you use it, or what capabilities it's going to give to people. That part, I think, is going to be more and more valuable.

Ian Silber: Because anybody can theoretically build anything, I think the people that have the sharpest thinking and the clearest point of view, that's going to resonate. I think the culture of things is still going to matter, and the care and craft. There is going to be a real opportunity to stand out, given that you'll have all these tools to build whatever you want. So human judgment is only going to get more important.

Nat Rubio-Licht: Yeah. So I've been thinking a lot lately about this idea that AI can allow us to automate a lot of things. It can do a lot of things for us. But I guess the question is, what should we allow it to do for us? Are there any parts of the design process that are happening today that might be a thing of the past in five years?

Ian Silber: Probably. Yeah. I generally try not to make too many assumptions about what's going to change because it's really hard to know. We're already able to automate parts of the design process. Before, when you had to point and click to do something, draw a button and type in the text, we've already seen this throughout the years. When I first really got into product design, we were using Photoshop, which was really not designed for software design. It was designed for editing photos, and we hacked it to do UI design. It was really tedious.

Ian Silber: Then tools like Sketch came out, and then later Figma, which optimized the workflows and made it really good for doing the things that you do specifically for designing software, and then made the collaboration really good. You can think about that as a spectrum of what used to take a long time now being easier and faster and more optimized. Now with AI, we're taking that to a whole other level.

Ian Silber: I was working, or a designer shared something with me just last night, where they had an idea and they were able to use Codex to design it. It was a very unique idea that came from this designer's head, that would have been hard to come from many other places. It came with a very unique perspective and point of view. He was just able to use Codex to do that. He could have done that manually before, but the fun part or the important part of the work wasn't necessarily how he built it, or having to specifically write or draw things, but the actual output is what mattered. He was able to do that in much less time. So parts of the process are always changing, but the output, I think, is still not really changing.

Nat Rubio-Licht: Gotcha. I've got one more question for you. You've given a little bit of advice already. What's a piece of advice you would give to young designers right now that may be graduating and are facing this new world of AI?

Ian Silber: I think just really have fun with it. Understand that it's all evolving very quickly. Like I said, things that might just barely work now, in a month maybe will work really amazingly, or maybe it takes a year. One thing I think a lot about is just how far things have come in the last couple of years. If you think about projecting out, especially as an early-career designer, how much has changed in my career, I can only imagine what's going to happen in the next 10, 20 years or whatever.

Ian Silber: So I think just being super curious about it. In general, I've always found that great designers, or great anything, great builders, approach their work with a curiosity and an excitement to try new things. I do think the lines are blurring, but I also think there's opportunity, like I said, to just be really good at something. There's always an opportunity to stand out if you really can obsess over something or get really, really good in a certain area.

Ian Silber: There are so many ways you can think about being an engineer or a designer or product manager. And yes, at the end of the day, we're all builders. We all want to build something. But I think AI can help you hopefully do more of what you love to do.

Nat Rubio-Licht: Yeah. Is there anything that you wish you knew when you started at OpenAI that you know now?

Ian Silber: Oh, that's a good question. I knew nothing about AI, honestly, which was maybe good. I don't know. It was all evolving quickly. I'll be honest, I remember joining and everybody was talking about agents, and I had no idea what that meant. I was like, what? This was like three years ago. It took me a while to even understand, because agents didn't really, they weren't really working. They were more of an idea in some ways, at least at the product level.

Ian Silber: Another way to think about this is we're still so early. I think approaching everything with a beginner's mindset and just knowing that you shouldn't feel like you're falling behind because, oh my god, I'm not up with the latest on agents. It's all changing so rapidly. It's very hard to keep up with it, even me internally, to understand exactly the latest breakthrough and everything. So I think that goes back to the curiosity thing, and just really trying to understand how it can help you or use it in your own workflows.

Nat Rubio-Licht: Yeah, being willing to constantly learn.

Ian Silber: Totally. Totally.

Nat Rubio-Licht: Well, I think that's all I've got for you. Thank you so much for taking the time to join me. This was lovely.

Ian Silber: Awesome. Thank you.

Nat Rubio-Licht: Thank you.

Ian Silber: All right.