Chatbots Aren't the Final Interface: OpenAI's Ian Silber on Designers, AI, and What Comes After the Chat Box
Lenny's Podcast
Ian Silber has led product design at OpenAI for three years, covering ChatGPT, Codex, and the company's broader product experience. Before that he worked at Artifact, the news app from Instagram's founders, and spent eight years at Instagram. In this conversation with Lenny Rachitsky, the central question is why designers feel so uneasy about AI right now, and what that unease misses. Silber's position is that the anxiety is rational but the moment is an unusually good one for designers. He also describes how OpenAI actually designs its products, and where he thinks ChatGPT and Codex are going as they move beyond a plain chat window.
Why designers are the unhappiest group in tech
Lenny opens with results from his ongoing sentiment survey of the tech workforce, which he describes as the largest survey of its kind. Designers and user researchers came out as the most unhappy group on every dimension he measured. They reported being the most overwhelmed, the most anxious, the least optimistic, the most worried, and the most tired. They were also the least likely to recommend their role to people entering the workforce.
Silber says he sees a mix. Many designers he talks to are optimistic, but he understands the uncertainty. People are asking whether the role is changing, whether expectations are different, and whether they now need to ship code every day. For someone "classically trained in a certain way of working," he says, the pressure can feel like "Oh my gosh, I need to really change everything." Engineers have already gone through their shift and are visibly working in a completely different way. For designers, he says, "we're unclear right now what is expected of a designer," and that lack of clarity breeds fear.
He points to a specific frustration that came up in OpenAI's own internal surveys. Engineers have seen "10 or sometimes like 100x" gains in productivity, but the design team hasn't. His explanation is about the nature of the work. With a coding agent, a request is fairly binary: you ask for something and it usually gets it right. The design process is still "very messy and fluid." You have an idea, try it, and it's bad. You try again and it's still bad. Then users or internal reviewers tell you it's wrong. AI has compressed part of this because you can generate far more ideas quickly, but the feedback loop remains, and so does the overhead of aligning many people in a large company.
How OpenAI tries to lower the pressure
Silber contrasts that general unease with what he sees in his own team, and he ties the difference partly to OpenAI's culture. The company started as a research lab and still treats itself as one. He calls that DNA "deeply embedded in everything we do," and says it is something he has come to love. When hiring, OpenAI looks for designers who don't necessarily have an AI background but are genuinely curious about the tools.
The team has found that supporting exploration, instead of demanding that people change how they work, reduces the pressure. Before the interview, Silber asked both very senior and early-career designers at OpenAI how they felt. He reports they were excited: they feel more able to be creative, to express themselves, and to try bad ideas and throw them out quickly. He says the bar on his team is rising and the thinking is getting stronger the more people use these tools. People who embrace this, he says, find they have much more agency and can have much higher impact as designers.
Lenny notes that his survey found a similar split. About half of respondents are having the best time of their careers and half are struggling. The strongest correlate of the positive group was feeling "amplified" by AI. Asked what else distinguishes designers who are thriving, Silber says they use AI at every step of the process. For example, instead of starting a new idea the usual way, they might drop it straight into Codex, Claude, or another agent to help think it through or prototype it. He suggests the broader design community could do more to help each other learn these new ways of working.
He puts heavy weight on adaptability, which he acknowledges has become a cliché but calls "so true." What didn't work yesterday may work today. He gives a personal example. On paternity leave toward the end of the previous year, he tried coding agents on side projects and kept hitting hiccups. About a month later Codex came out, he tried again, and "it just works." That kind of shift, he says, changes the design process, because now you can drop an idea in and simply try it.
"Nobody should feel behind right now"
A recurring theme is that everyone is early. Silber says it may feel as though the industry is already far into this transition, but "we're all figuring this out as we go," and nobody has a settled answer for how the design process should work. Whatever works today will probably change again. So, he argues, no one should feel behind: "if you literally started today, you're going to have a leg up on pretty much most people."
He illustrates how fast things move with his own time at OpenAI. When he joined three years ago, he got demos of an early voice mode that hadn't shipped and a computer-use feature that "barely worked." Both have now become part of how OpenAI designs and builds software. What looks cutting-edge today, he says, will look archaic in a year, "let alone a month."
Why he calls this the best time in history to be a designer
Lenny recalls that before the interview Silber told him this is the best time in history to be a designer. Silber stands by it. He describes learning design and code as a teenager by viewing the source of HTML pages and using tools like Dreamweaver. Compared with that, you can now go from idea to working version "in a matter of minutes," learn extremely fast, and be very expressive. "Anybody can be a designer," he says.
His deeper argument is about differentiation. As more software and products get made, he believes the human element of design becomes more important. Companies with great design have always been able to stand out, but when "anybody can make anything," products that fully understand their users and create something delightful will be judged on exactly that. That, he says, is the designer's core job.
He also sees designers becoming more empowered. He describes friends thinking about starting a company who were rethinking the designer-to-engineer ratio. Startups he has seen might have had one designer to about 15 engineers. These friends were instead considering two designers and one very strong engineer: the engineer to make sure the product is solid, and two creative people to shape what gets built. Inside OpenAI, he says, designers are "completely empowered" to take an idea or an early research concept and work out how to package it into something people want to use or that meets an unmet need. He suggests more people may start thinking of themselves as designers.
Lenny pushes back a little. If design is such a differentiator, why hasn't there been a visible surge in design hiring, like the two-to-one ratio? Silber doesn't claim it has happened. He says it may start to happen, or engineers may become more design-focused. He adds that designers are getting more productive, though not in the same direct way engineers have been.
PM, design, and engineering: blurring but not merging
Lenny brings up Marc Andreessen's description of the current moment as a three-way standoff. PMs, engineers, and designers each think the other roles will disappear and theirs will win, like the Spider-Man pointing meme. Silber's view is that the roles are definitely blurring, but that they always overlapped. He has seen designers who are really PM archetypes and designers who are essentially polish engineers. In his view, single siloed role definitions have never really matched reality. Big companies need career ladders to organize people, but real teams are made of people with different strengths playing different roles.
Still, he argues, especially from a larger-company perspective, distinct responsibilities remain valuable. It is hard to find one person who can design, rally a team, align the company, do all of a PM's work, and also engineer the product. He still values an excellent PM who can bring multiple teams together, sharpen the problem a team is focused on, and handle resourcing. He likewise values an engineer who makes sure the output is sound and fits the broader system. There may be more movement between roles, but he thinks of them as "different hats that you need to wear to build great software."
He qualifies this as not a blanket statement. If he were starting a company today, he would hire the best generalists he could find, people who can move fluidly between roles. He adds that this may always have been true for startups.
Is AI already a good product designer?
Asked whether AI will ever be an extremely good product designer, Silber answers that "it already is an incredible product designer," and he means it. What he finds notable is that it's accessible to everyone. But he immediately separates that from being the best at any one discipline. It isn't necessarily the best visual designer, or the best at information hierarchy, typography, user experience, or interaction design. He says it's impossible to predict where things go, though the models will certainly keep improving. He frames OpenAI's aim as building tools that let people be more creative and do and build more. The model is already a great designer, but it's up to the user to figure out how to get the best out of it.
On where human judgment will stay most valuable, he lists three things. The first is truly understanding what people need, including the feedback loop of watching people use something. The second is inventing something new. The third is the distinctly human point of view, the sense that someone made this and had a take on it. He believes that last quality will stand out more and more, and that designers should work on answering "what is my point of view? Who am I building this for and why?"
Lenny relays points from an earlier conversation with the PM of Codex about why AI hasn't been great at design, noting it has improved since then. One point was that great design often means doing something new, while a model trained on today's most interesting websites reproduces what is already no longer novel. The other was that AI researchers aren't necessarily strong designers, so they may not know what it takes to train a model to be good at design.
Silber picks up the novelty point. He argues that the most important products of the past 10 to 20 years involved designing for things with no precedent. The iPhone meant designing for multi-touch, which opened a new interaction paradigm, and "there's no training data for that." Instagram embraced the fact that everyone suddenly had a phone. Snapchat used interaction design some people considered strange but "flipped on its head" how people might want to communicate. For him, that is great design: truly understanding an unmet need and addressing it in a unique way. AI products are in the same situation now. How do you talk to an AI with your voice, or use an agent to carry out tasks? There isn't much precedent. Humans, he says, are good at judging what good looks like, or at least at trying many things and learning alongside users. He adds that this could change quickly, but sees many reasons why a human in the loop will be important.
What OpenAI looks for when hiring designers
Silber starts with a team-level principle. He believes in building a well-rounded team with different strengths, acknowledging this may be a luxury of scale but thinking it also applies to smaller companies. That means the right proportion of generalists who can do everything, alongside specialists. He names excellent visual designers, a brand team that shapes the brand's expression inside and outside the product, strong prototypers, more strategic product thinkers, and people obsessed with the cutting edge. He hasn't found a single designer who checks every box, and he thinks it's valuable to embrace what each person is naturally drawn to.
For individual candidates, he names several qualities:
- Curiosity rather than an AI background. He tells candidates OpenAI doesn't expect AI experience, because the field is moving too fast. It does expect strong curiosity and aptitude for learning how the technology works and how to use it in their process. At OpenAI especially, it also expects them to think about how to build tools that let more people use it.
- Prototyping. He has always valued it, and it has become far more accessible, so it tends to stand out in candidates.
- A point of view and strategic thinking about where these tools are going.
- Systems thinking, which he says is becoming more and more important.
Systems thinking and "just do less"
Silber explains why systems thinking matters in OpenAI's specific context. The company is building what he calls "the super intelligence that has all these different capabilities," so designers need to think about underlying primitives that build on each other. He cites Notion as a product that did this well early on, with composable building blocks. OpenAI doesn't want to build one-off, siloed experiences. It wants to ship systems that stack, so that the product feels like "one cohesive simple thing" and the model can reason about how to combine its parts. Given how complex the range of possible actions is, he says, understanding the system becomes even more important.
Lenny notes this is the fifth podcast in a row where systems thinking came up, and connects it to designers' stress. There is less time for the traditional process when engineers are shipping huge numbers of PRs. Silber responds with counterintuitive advice he has been giving his team: "just do less." Don't design something if you don't have to. There may already be a component or system to build on. The feature may not be needed at all, and extending an existing one might be enough. Part of this is thinking beyond the bounds of your own project and across the whole product, which he says matters more now because of the speed.
Balancing speed and craft
Lenny cites his conversation with Jenny Wen, head of design at Claude. She argued that the traditional sequence of prototype, mock, test, research, iterate, and hand off to engineering no longer has time to happen, and that the role is shifting toward big-picture direction and steering. Silber says OpenAI "100%" works that way sometimes, but it depends on the stage of the work and how mature the product is.
Some ChatGPT features get obsessive attention: "we try 100 things, we throw out 99, we finally ship one." He gives the composer as an example of something that changes daily, with many attempts thrown away. For those features the team sweats the details, runs A/B tests, does user research, and iterates heavily. Other work, such as Codex and the super-app effort, involves trying things very quickly and "building in public." He says that is a different way of working, but once you get comfortable with it, the instant feedback from inside and outside the company is satisfying.
The main discipline, he says, is accepting that the technology may change within a month. An interaction or setting may no longer be needed, a new one may become necessary, the latency profile may change, or the intelligence may improve. So designers shouldn't be too precious about something until it has "hardened" to the point where it probably won't change tomorrow because of the technology. At that point, they make it the best it can possibly be.
As an example of the slow path, he describes work on how ChatGPT should feel as it becomes deeply embedded in people's personal lives. That won't be designed and shipped overnight. The team is trying many ideas on whiteboards and in Figma, prototyping, and dogfooding different versions long before shipping. At the other extreme, some things go "idea to ship in like, you know, 4 hours." Lenny summarizes this as picking your battles and focusing design effort on what will be durable. Silber agrees and adds, laughing, that "it's hard to know what that is."
Designing for a billion users and for power users
Joel Lewenstein, head of design at Anthropic and a friend of Silber's, suggested a question for the interview. He imagined the hardest part of Silber's job is balancing the needs of ChatGPT's roughly billion monthly users against the need to keep inventing. Silber says he has never worked with such a spread of audiences. Instagram also had billions of users, but the product was fairly simple and could be used in only a few ways, with some special consideration for creators and large accounts. ChatGPT serves someone deciding on tonight's recipe or brainstorming a haircut. It also serves someone automating their farm in Japan. "It's just like the broadest spectrum ever."
OpenAI works from the idea of a "capability overhang": the vast majority of people get only a sliver of the value the model could offer. Silber says that isn't necessarily bad, since they still get a lot of value. The approach is to keep an extremely simple, streamlined experience for most people while making cutting-edge capabilities accessible to those who care. The risk is over-indexing and putting something confusing in front of a billion users. So OpenAI experiments early with enthusiasts through the desktop app and Codex, and now ChatGPT Work, which he says is stepping into a much broader audience. He believes something only truly reaches billions of users when they don't have to think about a switch or a mode. Eventually, new capabilities have to be distilled into the main experience. Designers have to hold empathy for everyone from free daily users to people paying hundreds of dollars a month, which he says adds complexity but also makes the work fun.
The blank-box problem and moving beyond pure chat
Lenny observes that few products span a range from "what's this rash on my finger" to building something as complex as Salesforce. Silber connects this to the "blank box problem" the team constantly thinks about. Critics ask whether chat is really the final frontier or just a "glorified terminal." The design challenge is making something that can "shapeshift into anything," whether someone arrives to generate images for fun or to produce a PowerPoint. OpenAI wants to help novices understand how it can add value to their lives or work, and to get out of the way of people who use it as a daily workhorse.
Lenny recalls former OpenAI CPO Kevin Weil's point that talking is a good interface for something whose intelligence keeps rising, since humans talk to each other across a wide range of intelligence. Silber agrees and describes two design directions. The first is helping users express intent better without having to become expert prompt engineers. That includes teaching them what the model can help with and making sure it understands what they want. He mentions a recently shipped feature where an open-ended question can trigger follow-up questions the user answers by tapping instead of typing.
The second is that the transcript no longer needs to be pure text. He describes "writing blocks." When you ask for a letter or email, the result appears in a block you can manipulate directly: type into it, brainstorm, edit parts of it, or copy just that portion. He says this has made that interaction much better, and similar special tools are being built for image generation. He expects the UI to keep evolving toward context-aware affordances. ChatGPT might look and behave differently for a designer, a data scientist, or someone using it to run their home life.
Where ChatGPT and Codex are heading
Asked about the long-term vision, Silber names several directions. First, ChatGPT will become more adaptive to the full context of your life. Second, proactivity, which he calls a big untapped part of the experience. He says ChatGPT Work already offers fairly good suggestions based on sources like your calendar or Slack. Third, voice: he says a recent step change in the model makes it feel much more natural, and he can only imagine what it will be like in two or three years. Fourth, a shift from answering questions to actually doing things for you.
He hopes all of this gets simpler. For the broad consumer base, the super-app vision is one universal input that knows whether to answer quickly, be chatty, or go off and do work. Users shouldn't have to think about which model or speed they're using. Those options should still be exposed to people who want them, but the default should work well for most people. He also expects richer voice input, richer and more interactive visual outputs, and durable workflows you can reuse. Right now, he says, it often feels like starting from scratch each time.
What he wishes he'd known on day one
Silber admits he was quite intimidated joining OpenAI. GPT-4 had recently come out, and while he was using the products, he didn't feel he truly understood them. He says he still has a lot to learn. The first thing he'd tell his earlier self is that it's early and everyone is figuring it out.
The second is more personal. He had managed before, but never at this scale, on a team growing this fast, at a company this different, where every day brought something new. For a long time he felt he was bad at the job and didn't know what he was doing. When he talked to friends who had led large design teams or were going through similar transitions, he found they felt the same way and his problems weren't unique. His advice is that imposter syndrome shows up at every career stage. Finding an impartial mentor or peer outside your situation who is facing similar challenges helped him a lot.
Lenny brings up Elena Verna's post "Please Stop the AI Confidence Theater," about people pretending on social media that they've figured AI out. Silber prefers humility: admit that it's changing quickly and embrace it. He agrees with Kevin Weil's line that today's model is the worst the model will ever be. He compares the current moment to designing for early computers, when components evolved quickly and limitations disappeared in sudden step changes. What was completely broken yesterday "now actually just kind of works."
Advice for overwhelmed designers
For designers who feel anxious and uncertain, Silber first says the feeling is "totally rational." He then encourages treating this as a time to explore. Try working differently, rethink your tooling, prototype more, and have fun with it, remembering that everyone is still early. His second point is to focus on outcomes. "We're not here to build processes" or "token max." The goal is to make great software and products people love. If you focus on that, he says, you'll likely find these tools help, and you still have a large role as the person steering and directing them.
How he uses AI in his own work
Silber says he spends less time "in the pixels" now. He uses AI to think through broader ideas and to visualize early concepts. If an idea comes to him at night, he types it in, has it prototype something, and brings the result to his team. He finds ChatGPT Work and Codex useful for this early, exploratory part of the process. He also uses AI for operations: summarizing important Slack messages he needs to follow up on, preparing context before important meetings, and supporting recruiting.
Asked how he actually does this, he highlights ChatGPT Work, which he says now runs fully in the cloud. He can be anywhere, describe what he wants, and let it spend time and come back with a good artifact. The fact that he doesn't need to be near a computer is part of the appeal.
Failure: IGTV, Reels, and learning on the go
For the failure corner, Silber first admits that at OpenAI he felt like he was "kind of failing every day" early on, not fully understanding how to build a design team or how to hire, and adjusting quickly. He acknowledges this may sound like a cop-out, but says his whole career has worked that way, without much long-range planning.
His concrete example is IGTV at Instagram, which he calls "a huge flop." The team had baked in some wrong assumptions and held on to certain constraints. They learned quickly, changed those, and rebuilt, and the result was Reels, which became very successful. He says experiences like that have hardened him to critique. What matters more than any single launch, he argues, is how you react, fix forward, keep iterating, and listen to users.
In closing remarks, he says he hopes designers can stay adaptive and focus on making great things. He believes great companies are seeking great design, and that some companies will be left behind because they haven't shifted, but there is "plenty of room for amazing, great design."
Lightning round and lessons from Groupon
Silber recommends The Design of Everyday Things, calling it a classic that shapes how he thinks about design. For entertainment, he picks the Broadway show Maybe Happy Ending, which he notes hit close to home because it's about super-intelligent humanoid robots, and praises its production and set design. He also recommends the film The Invite, which he enjoyed for its simplicity: essentially four people in one apartment, with strong writing and acting. His favorite recent product is his Rivian. He appreciates how it discards old assumptions about operating a car: you walk up, it unlocks, and you don't turn it on or off. He says that if he could own a Waymo, that would be "the most magical experience in the world." He doesn't have a life motto, more a "vibe": have fun, try things, don't feel you have to fit in a box, and stay flexible.
The final question is what he learned at Groupon that helps him now. He credits then-CEO Andrew Mason with giving the product real character. Groupon had a writing team made up mostly of comedians, and in its early years the product didn't take itself seriously. People loved the email not just for 50% discounts but because it was well written and funny. The broader lesson Silber carries forward is that every company has distinct DNA and you have to adapt how you work to it. Groupon worked one way. Instagram, under Kevin Systrom, was design- and creativity-focused, with a CEO who worked with designers as a designer himself. OpenAI, under Sam Altman and rooted in research, is different again.
Lenny adds that you usually can't change a company's culture, especially a founder-driven one. Silber frames his approach as helping to "round it out" rather than change it. Instagram moved more slowly and changed the product less, which he says worked well for most of its stages but wouldn't suit OpenAI right now. He brought Instagram's motto, "do the simple thing first," to OpenAI and still uses it, while acknowledging it isn't always the right approach, since OpenAI sometimes ships things that are still experiments.
Finally, Silber asks designers, engineers, and PMs to share how their ways of working are changing. He says the design process is becoming more opaque, and showing it openly is "such a gift to the community." He can be reached on Twitter, and his personal website includes a portfolio of past work and a hidden game whose high scores, he jokes, are out of control.
I've been doing this sentiment survey of the tech workforce. Surprisingly, designers are just the most unhappy on every dimension.
We're unclear right now what is expected of a designer. Someone who might be classically trained in a certain way of working now feeling like, "Oh my gosh, I need to really change everything."
You believe this is the best time in history to be a designer.
We're all extremely early in this process. There's just a huge opportunity for designers. Nobody should feel behind right now. I think that if you literally started today, you're going to have a leg up on pretty much most people. There's just so much opportunity to change the way you work to make great things.
If you look at engineering, it's just a completely new job versus a few years ago. I'm curious how that translates to product design.
Engineers have 10 or sometimes like 100x their productivity, but our design team hasn't because the design process still takes time. It's still sometimes you have to try a bunch of things and throw them out. Still very messy and fluid.
We had a podcast with the head of design at Claude, Jenny Wen, where she talked about there's been this traditional design process: prototype, mock, test, research, iterate, and now there's just no time for that.
There's certain features for ChatGPT where we really obsess and we really spend a lot of time. We try 100 things, we throw out 99, we finally ship one. And other things we really embrace building in public. We take big swings and then we learn quickly.
Do you think AI will ever be an extremely good product designer?
I think it already is an incredible product designer.
Today my guest is Ian Silber, head of product design at OpenAI. Ian's been in this role for three years now, which is a lifetime in AI time. Prior to OpenAI, he was at Artifact, the news app from the founders of Instagram that I absolutely loved. Before that, he spent 8 years at Instagram. The role of design is in such an interesting place right now, and Ian has such a unique and insightful and fascinating take on the future of product design and what the AI tools that we use every day will look like in the future. A huge thank you to Cleo and Joel Lewenstein for suggesting topics for this conversation. Before we get into it, don't forget to check out lennysproductpass.com for a year free of the hottest and most beautifully crafted AI products in the world, available exclusively to Lenny's Newsletter subscribers. With that, I bring you Ian Silber.
Ian, thank you so much for being here.
I'm so excited to be here. Thanks for having me.
I'm even more excited. Something that I'm very excited to get your take on. I don't know if you've watched me doing this, but I've been doing this sentiment survey of the tech workforce over the past couple years. I think it's the biggest survey of its kind, where we basically survey people in tech about how they're feeling about their careers, their work, their roles, especially with the rise of AI this last year. And kind of surprisingly, designers and user researchers are just the most unhappy on every dimension, like every actual dimension. They're the most overwhelmed, the most anxious, the least optimistic, the most worried, the most tired, also the least likely to recommend their role to people entering the workforce. I'm curious if you're seeing this yourself in the design community, and if you are, what do you think is going on?
First of all, it's an incredible survey, and it's such great, I think, information to have as an industry. And I think it's something we need to unpack, you know, and I see a mix for sure. I think I can speak most to design. And I mean, I talk to a lot of designers that are very optimistic and very excited. And so in my world, luckily, I see a lot of that. But I can totally understand why people might be feeling very uncertain, right? Because there's so much discussion about, well, is the role changing? Should my expectations be different? And, you know, well, do I need to do X, Y, and Z? Do I need to be shipping code every day? And you know, someone who might be sort of classically trained in a certain way of working now feeling like, oh my gosh, I need to like really change everything.
And I think, to me right now, it feels like the uncertainty of that I can totally understand, right? Because as an engineer, I think people have already sort of gone through that shift. Like we see that engineers are working totally different. And we also, you know, one thing I do hear, and this came up a lot in my own team when we did, we do similar surveys internally. One of the big things that came up was, you know, we see that engineers have 10 or sometimes like 100x their productivity, but our design team hasn't, right? Because the design process still takes time. It's still sometimes you have to try a bunch of things and throw them out, right? Versus maybe working with a coding agent, you ask it to do something, it's pretty binary now at this point. It usually sort of gets it pretty well.
But the design process is really still very messy and fluid, and you think you have a great idea and you try it and it sucks, and then you try it again and it sucks, and then you give it to, you know, users, or you get internal feedback and it's wrong, and you still have to go through that arduous and kind of messy but really fun process. And that hasn't really... it's compressed in that you can get many more ideas out fast, but you still need that whole sort of feedback loop. There's the alignment of getting many people, you know, especially in a bigger company, like that still has a lot of overhead. So the point is that I think we're unclear right now what is expected of a designer, and that lack of clarity I can see totally lead to, you know, some fears or some unease, I guess.
But on the other hand, I've seen that, I mean, our designers... what's been great, I think, about OpenAI is we're such an experimental company. We are driven by, you know, research, and we treat our company as a research lab, and we totally embrace that. And that's a new thing that I really kind of learned and have really come to love about OpenAI, is that it started as a research lab, and that DNA is deeply embedded in everything we do. And as designers, I think that, you know, you come here, we look for people that don't have a background in AI but are really curious about the tools and how it can... And we found that just by sort of supporting that excitement to explore, people feel less pressure. It's not like you need to change the way you work, but it's more about encouraging exploration.
And when we've done that, when I go and talk to both very senior designers and early career designers at our company, they're so excited because... I literally was just asking some people before this 'cause I wanted to get some different takes, and, you know, I think the feeling is that they have so much more ability to be creative, to express themselves, to try bad ideas and throw them out. You can just really move through lots of different ideas, and it becomes this incredible way to create. And the things that I'm seeing from my team, the bar is raising. I think the thinking is getting stronger the more that they're using these tools. So I think for the people that are starting to embrace this, they're starting to feel really excited and realizing that they have just so much agency as designers and, you know, so much more ability to have really high impact as a designer.
This episode is brought to you by our season's presenting sponsor, WorkOS. What do OpenAI, Anthropic, Cursor, Replit, Sierra, Clay, and hundreds of other winning companies all have in common? They are all powered by WorkOS. If you're building a product for the enterprise, you've felt the pain of integrating single sign-on, SCIM, RBAC, audit logs, and other features required by large companies. WorkOS turns those deal blockers into drop-in APIs with a modern developer platform built specifically for B2B SaaS. Literally, every startup that I'm an investor in that starts to expand upmarket ends up working with WorkOS. And that's because they are the best. Whether you are a seed-stage startup trying to land your first enterprise customer or a unicorn expanding globally, WorkOS is the fastest path to becoming enterprise ready and unblocking growth. It's essentially Stripe for enterprise features. Visit workos.com to get started, or just hit up their Slack where they have actual engineers waiting to answer your questions. WorkOS allows you to build faster with delightful APIs, comprehensive docs, and a smooth developer experience. Go to workos.com to make your app enterprise ready today.
What I'd love to do in this conversation is help folks that are in that bucket of, well, this sucks and things aren't going well. Because as you said, and this is exactly what we found in that survey, about half of people are having the best time of their career, having the most fun, the most optimistic about where things are going, and then the other half are having a really hard time. And exactly what you said, the people that are having a great time report, like the strongest correlation is they feel amplified by AI. It's actually making them better. And that's the way you described engineers. They're seeing that now. They're like, "Holy, I can do so much more." And so just to kind of reinforce the point you're making, attributes of people that are doing well: feel amplified, very curious, have time to explore, feel like they can be more creative. Is there anything more along those lines of just patterns of people that are doing really well and having a great time that might be helpful for people to hear that are not having a good time?
What I see in the folks that I think are really embracing this, right, they are using it every step of the process to kind of help them, right? And I think it helps, you know, we have a community of course of designers that are all trying this together. So if there's any way to kind of find that, and that's something that we should think about more as a broader community, how we kind of help each other. But I think that learning kind of different ways of working, right? And so it might be, okay, I have this idea, let me just actually start by trying, what if I just threw this right into Codex or Claude or whatever tool you use, like throw it into an agent to help you think about it, to help you prototype it. Just try a different way of working. It doesn't necessarily mean that that has to be the exact right way.
But I don't know, I think we found that people that are embracing this are adaptable. I think adaptability is another term that I think we all hear a lot now, but it's so true. It's knowing that this stuff changes very quickly. What didn't work yesterday might work today. I mean, literally within, you know, six months this entire industry has shifted. I remember I was actually out on paternity leave towards the end of last year and I was working on some side projects. I was trying some coding agents and it worked, but it just kept kind of running into these hiccups. And then like a month later, Codex drops and I try it again and it's like, oh my gosh, it just works. And that I think changes a lot of the design process, because now all of a sudden if you do have an idea, you can plop it in there and just try it. And so anyways, I think it's just really about not being scared.
I think another thing to think about is that we're so early. We're so in the early innings of this entire evolution, right? It feels like things have changed a lot, and it feels like maybe we're pretty far into this shift. But the good news for anybody who is trying to design or get into design, whether you're early career or late career, or you never even actually designed anything but you've been thinking about it, is we're all figuring this out as we go. Nobody has some perfect answer about how the design process should work. And whatever you think is working today will probably change again. And so I think what's encouraging about this is that nobody should feel behind right now. I think that if you literally started today, you're going to have a leg up on pretty much most people. And just really thinking about how you can just make great stuff. And there's just so much opportunity to change the way you work, or to just make great things.
And what we think is really cutting edge today will look so archaic in, you know, a year, let alone a month, right? And that's something I've learned. You know, I've been at OpenAI for three years, and I think about the stuff that we were working on when I first joined. I mean, I remember getting a demo of our early voice mode that we hadn't shipped, or computer use that, like, you know, barely worked. And now it's literally become part of the process of how we design. We use both of those tools to build software. And so anyways, the point is it's changing very quickly. We're all extremely early in this process, and nobody really has it figured out.
I love that. Also, when we were chatting ahead of this, you actually told me you believe this is the best time in history to be a designer.
Yeah, I mean, I truly believe this. I can't even imagine, you know... So, okay, history lesson: when I was getting into design, I was in my teenage years, right? And to learn things, you know, I was kind of learning design and code, and you could view source on an HTML web page and kind of see how things were built, and that was kind of how you had to learn. And, you know, I don't know, there was Dreamweaver, and there were these tools that actually were really cool. But to compare those to today, if I were just getting into design, the fact that you can learn so quickly, you can go from an idea to a working version of it in a matter of minutes. You can be so expressive, and I think anybody can be a designer, right?
And I think that as more and more software is made, more and more products are made, I do think that the human element of design is going to really start to become more and more important. That's always been the case, I think, that, you know, companies with great design can of course stand out. I think it's going to be more and more important, right? I think that products that fully understand their users and that make something delightful, when anybody can make anything, you know, that's going to become one of the most important dimensions of a product, and that's how people are going to evaluate it. And so I think that is our core job as designers, right?
And then on top of that, I think that designers are just much more empowered. So, you know, I was talking to some friends that are thinking about starting a company, and talking about the ratio of design to engineering, and how at a startup maybe that's totally different. It may have been like you're going to have one designer and, I don't know, 15 engineers or something like that. You know, at a startup I've seen that sort of ratio. But they were talking about, well, maybe we'll have like two designers and one engineer. One really good engineer who can make sure that we're really building something that's rock solid from an engineering standpoint. But then we want two really creative people that are going to be thinking about how we kind of build this product. And so thinking about that, I
think that there's just like a huge opportunity for designers. I think that internally we see that more and more we're, we like designers, we are totally completely empowered to take an idea, to take an early research concept and try to figure out how we can actually like package that up into something that people want to use or that solves some sort of unmet need for our customers. And that is a designer's job. And so I think that maybe more and more people start to think of themselves as designers. But either way, I think it's like the best, I really do think it's the best time to be a designer.
So along those lines, when Marc Andreessen was on the podcast, he had this hilarious way of describing the state of the product org right now. He's like basically there's this three-way standoff. PMs, engineers, designers, they all think the other roles are going to go away and they're the ones that are going to win in the end. And so there's this like Spider-Man meme where they're all pointing at each other. And I'm curious because I can see a case for each role becoming the dominant role. And I completely agree with you that with all of the tools being shipped, like there's like a thousandx explosion in products being shipped, there's just like endless streams of things to pay attention to. You'd think design would be such an opportunity to differentiate and stand out. I haven't seen that yet in just like an acceleration of design hires and explosion as you said, like two designers to one engineer. Do you think that's coming? Why do you think that hasn't happened yet? Do you think it's, as you said, just early and people will realize, oh wow, we need a lot more designers? What do you think is going on?
We might start to see that happen more. Or if anything, we might start to see engineers wanting to become a little bit more design focused. I also think that designers are getting more productive. It's not necessarily the same direct output of like engineers 10xing. But we are seeing that designers are able to do a lot more. But no, I mean I think that that is something we're sort of always thinking about and evolving.
The other thing I wanted to talk about too is like the difference between PM and engineering and design, and are they all blurring together? And my point of view is that they definitely are, and there's always been overlap as far as I've seen it, right? I've seen designers that are like kind of really good product, you know, manager type archetype, but can design, and I've seen designers that are like really good sort of polish engineer type, you know. So I think I've always found that these like single siloed role definitions don't really make sense. It's just that at big companies you have to have like a ladder to put people on so you can generally kind of like organize. But in reality, you know, when you look at a team, you see that there are people that have various strengths and people that play different roles. And so I've always kind of seen that. I think those lines have always been blurry, but they're becoming blurrier for sure.
But the other thing is I do still think that, and maybe this is thinking about it from like a slightly larger company perspective, but I think you do still need people that have certain responsibilities, right? Because if you have one person that's expected to design, to rally a team around something, you know, to align a whole company, to do all the work that like a PM does, for example, and also be designing it and also be engineering it, it might be hard to find one person that's good at all of those different skills. And so I still find a lot of value, for example, in having like an excellent PM that can help rally multiple teams together, that can help sharpen kind of the perspective or the problem space that a team is focused on, that can help with like resourcing, all of those things, right? And so I do think there will still be value in these and people focused on different parts of building a project or a product. That doesn't mean that maybe there's more fluidity, that people can move between them, but having somebody who is sort of generally responsible for these different things, or an engineer who's like really making sure that the output is totally sound and, you know, is actually like laddering up to something that is going to work for the broader system.
I completely agree. So just to make sure I understand what you're saying: there's talk about all the roles blurring, like maybe we'll all just be builders and we do everything, and you're saying, like, no, that's not going to happen. I'm picturing Venn diagrams of product, design, engineering like converging, but you're saying they're going to continue to stay their own roles. There's going to be a PM, design, engineer, but each can do more of the other roles.
I think that's right. I do think that like, and again this is not a blanket statement, right, because I think if you're at a startup that's like very different. If I were starting a company right now, I would hire like the best generalist people I could possibly think of, someone who I think could kind of fluidly move between that. But maybe again, maybe that's always kind of been true. But at a bigger company, I do think that I've seen like designers that are, you know, maybe incredible at designing, but could really use a product manager who can help them kind of hone both their thinking and then how they work and the operations of it all. So the point is I do just think that, it's almost like, taking the people out of it, I think that there are different sort of hats that you need to wear to build great software, and I think that those continue to be important functions, I think.
So this begs the question, do you think AI will ever be an extremely good product designer? And it's interesting to ask you this question because on the one hand you're like, oh yeah, ChatGPT is going to become incredible at everything. On the other hand, it's hard for a head of design to say, "Oh, no, design's going to be done by AI." I'm curious to get your take, and you know, it like sounds crazy, but like engineers never thought it would be able to write code the way it is now. So just, yeah, thoughts on just how good AI will get.
Maybe one way to put it is I think it already is an incredible product designer. [laughter] It really is actually already great. The cool thing about it though is it's accessible to every single person. And it's a tool that anybody can use. I do think it's really good. Now, that doesn't necessarily mean that it's like the best visual designer or best at, you know, information hierarchy or typography or even, you know, user experience and interaction design. It's not necessarily like the best at any of those. It's impossible to predict where things are going. And like I said earlier, like, you know, how much has changed in three years? Of course, they're going to continue to get better. But the way we think about it, and I really do believe this, is like we're building tools to enable people to be more creative, to do more, to build more. And so it is already a great designer, but it's at your fingertips to kind of figure out how to use it and how to get the best out of it.
Where do you think human brains will continue to be most valuable as AI gets better at product design specifically? What are kind of like the remaining areas for us to be useful?
Truly understanding what people need. You know, I mean the human feedback loop of watching people use something and understanding that. Inventing something new, I think. And then of course there's just the pure like human kind of take, right? Like, oh, somebody made this, somebody had a point of view about this. I just think that will stand out more and more, and I think that that's like an important thing for a designer to like really hone: what is my point of view? Who am I building this for and why? And I don't know, I think as AI becomes better and better, for sure there will be an opportunity to always kind of stand out and push, how can you push that even further.
Yeah, that touches on a conversation I had with the PM of Codex. I asked him just why is AI not good at design? It's actually gotten better since that conversation, which was just like a month or two ago. And he had a number of reasons for it. One is just like part of great product design is doing something new. And so if it's trained on all the most interesting websites today, it's just like, okay, but that's not interesting anymore. And the other point he made is like AI researchers aren't innately great at design, and so they may not understand exactly what it takes to train a model to be really good at this. Thoughts? What do you think? What do you think it'll take to get AI to be much, much better at product design?
I mean, the new thing is interesting, right? Because I think that everything we're doing is trying to design something that hasn't existed before, right? You know, if you look at great products that have come out in the last, you know, 10, 20 years or whatever, right? It's like all of a sudden, you know, take the iPhone, we're designing for multi-touch. And that opened up like this whole new interaction paradigm. And there's no training data for that, right? You're kind of really trying to understand this new thing, right? And then you had something like Instagram that was like really embracing the fact that everybody had mobile phones. And then Snapchat, I think, is another good example of great design, in that it was kind of maybe what people thought was strange sort of interaction design, but it completely flipped on its head the way people might want to communicate with each other. And to me that is like great design. It's like truly understanding, you know, like people have this problem or there's this unmet need, and I'm going to build it in this unique way.
So yeah, I do think that there's that sort of like fresh new kind of way. And then now with AI, it's like we're designing, like, how do you talk to AI with your voice, you know, or how do you use an agent to like carry out tasks for you? It's like, yeah, there isn't a lot of precedent for this. And so I think that, you know, humans are able to kind of do a really good job kind of evaluating, you know, what good is, or at least kind of like trying a bunch of things and learning along the way with people. And so I don't know, but again, like all this stuff changes really quickly. It's hard to say, but I think there's a lot of reasons why having that human in the loop is going to be so important.
And kind of following that thread of the human in the loop, when you're hiring designers these days, what's kind of trending up in what you look for, what you think is more and more important in the designers that will be successful today, especially at OpenAI? What's trending down, if anything?
This might be a luxury of having a slightly larger team, but if you're, you know, at our scale, you know, but even at a startup or a smaller company, I think this is still true. I really believe in hiring a well-rounded team that has many different strengths. So I think it's important to have kind of like your handful of, or the right ratio of, sort of generalists that can kind of do everything, but, you know, we also hire like incredible visual designers, or we have like a really amazing sort of brand team that helps us with kind of like the expression of our brand both out and in the product, or great prototypers. And so I think rounding that out, or great sort of product thinkers that are a little bit more strategic, or people that really obsess over like the very cutting edge of where these things are going. And I haven't found that one designer necessarily checks every single one of those boxes. I think people just naturally are, you know, drawn to certain things, and I think embracing that when you build a team, if you have the luxury of doing that, is like really valuable.
Okay. So that's like my broad statement. But then as far as what we look for, you know, I always tell candidates, we do not expect you to have a background in AI because, again, this is evolving so quickly, but we definitely expect high curiosity about it and a high aptitude for jumping in and learning how you're going to use this in your process, how it works. And not just how you use it in your process, but, especially for us, how you're going to create tools to enable more people to use it. And so it's more about the curiosity, I think, is something that we really push on or look for.
And then definitely prototyping. Prototyping is something I've always valued in my career. But I just think that, you know, it's become more and more accessible, and so that's something that I think really does stand out usually when we talk to candidates. And then I do think that, yeah, just like having a point of view on where these things are going, or just like any of the tools that people have worked on, and really thinking, that strategic thinking.
I guess last one on this is like systems thinking is becoming more and more important. And, you know, especially in our world, it's interesting because we're trying to build, you know, we're building the superintelligence that has all these different capabilities, and we need to think more about the underlying primitives that we can build that can kind of build on top of each other, right? You know, this is something that I think like Notion did really well early on, was like, we're going to build these building blocks, right? And then like everything kind of is composable. I think that's becoming really important too as you think about, like, well, we don't want to build all these sort of one-off siloed experiences. We want to be able to build and ship systems that can build on top of each other so that, you know, when someone goes and uses our product, it feels like one cohesive simple thing, and the model can reason about how it can pull all these things together. So that idea of really understanding the system and thinking about that is becoming even more important, especially because it's kind of complex, all the things that it could possibly do.
This is the fifth podcast in a row I've done where systems thinking has come up as a skill that people are more and more looking for, and it's like so funny. I need a bingo card for systems thinking, but it makes so much sense, and I think it especially makes sense from the designer perspective. If we come back to my take on why designers are kind of frustrated and stressed right now, it's just there's less time to do the traditional design process. Things are just moving, like engineers are shipping 5,000 PRs a day. And so to your point, this idea of building blocks and design system language, and just a platform for people to build on, feels incredibly important.
Totally. It's very hard, but yeah, it's like, it's funny, we've had these conversations where I'm like, just do less. [laughter] Like my encouragement has been do less. Like, don't design it if you don't have to. Maybe there's already an existing system or component or something that we can build on top of, or maybe we don't even need this feature. We actually just need to kind of, like, you know, extend an existing feature. I think that's an important thing, and part of that is like
thinking outside of the bounds of what you're working on, right? And thinking across everything I think is becoming much more important, and the speed for sure I think is a reason why that's so important right now.
Just to follow that thread, because that feels like a pretty big change in the design world. We had a podcast with Jenny Wen, she was head of design at Claude, and she talked about this, and I think it blew a lot of people's minds. There's been this traditional design process: prototype, mock, test, research, iterate, okay cool, ready for engineering, build it, iterate, ship it. And now there's just no time for that. And so the role in her eyes has just moved to big picture planning and just steering people in a good direction. Does that resonate? How much has that changed in the world of a designer?
I think that 100% we do work that way. It depends. I think for us it depends on what stage you're at in the process, or maybe how far along the product is. There are certain features for ChatGPT where we really obsess and we really spend a lot of time. We try a million things and we throw out... we try 100 things, we throw out 99. We finally ship one. The composer is changing every single day, and we try many things, we throw them out, and you could take that for many different things within ChatGPT.
And then there are other things like Codex where, I don't know, I mean the super app is a good example, where we definitely are just trying stuff really quick. I think that we really embrace building in public, which is a different way of working. But I think that when you get comfortable with it, it's actually really gratifying or satisfying, because you can try stuff and get instant feedback both internally and externally.
But yeah, I mean it is moving really fast, and I think it just depends, right? Like I said, certain things we really sweat the details, and we do test, we A/B test, we'll do user research, and we kind of iterate and evolve a lot. And then other things we take big swings and then we learn quickly.
But yeah, I don't know, I think the biggest thing is embracing that the technology might change in a month, and maybe a certain interaction or setting or whatever that you needed doesn't need to exist, or now a new one does need to exist, or a new capability is there, or the latency profile has changed such that you can rethink this. All these capabilities are changing, or the intelligence has changed. And so you have to be just really quick on your feet and maybe not be so precious about it until it's at the point where, okay, this is kind of hardened, we don't think that this is necessarily going to change tomorrow because of the technology, so let's make this the best it can possibly be.
We're thinking a lot about how we should evolve, really think about ChatGPT becoming deeply embedded in your personal life, and how should the app feel. That's something that we're not going to necessarily design and ship overnight and just put it out there. We're really thinking about that, and we're trying a bunch of different ideas, and we're putting it up on a whiteboard, or we're doing some design in Figma. We're prototyping it. We're probably going to dogfood a bunch of different ideas long before we ever ship something. But that's a bigger project, right? And then there's other things where we're idea to ship in, you know, 4 hours.
This all makes sense. Essentially, what I'm hearing is you pick your battles and you focus the design effort on the things that will be durable and not change.
Exactly. Yeah. And it's hard to know what that is.
Yeah.
This episode is brought to you by Mercury. Radically different banking loved by over 300,000 entrepreneurs, and now with Command. I've been a customer of Mercury's for over six years. I have never once thought about leaving. Mercury is basically what happens when banking is built by product people, not by bankers. They make it so easy, dare I say fun, to send invoices, move money around, set up virtual cards for folks on my team. Does your bank have an API, a terminal-native CLI, or an AI-ready MCP server? I don't think so.
And just recently, they launched Command, a conversational interface built directly into Mercury, which acts as your financial operator. I've been using Command to transfer money around, to figure out what categories I've been spending the most money in, analyze my cash flows, and just today, I used it to find out how much I've made from a specific sponsor over the past year. I just asked, "How much have I made from X over the past year?" 10 seconds later, I have an answer. It is so freaking cool. Visit mercury.com to learn more and apply online in minutes. Mercury is a fintech company, not an FDIC-insured bank. Banking services provided through Choice Financial Group and Column N.A., members FDIC.
So, I asked a bunch of people that know you well what to talk to you about. And Joel Lewenstein, who happens to be the head of design at Anthropic, is a good friend of yours, which I love. And he said what he imagines the hardest part of your job is, is balancing the needs of the existing billion monthly active users of ChatGPT with the need to innovate and push and iterate and invent. And so I'm just curious how you think about that, how you find that balance, knowing you have to cater to both sides.
That's a great question, and it's a really fun part of the way we work. I think that we have vastly different audiences, really, really different. I've never worked in quite such a spread. At Instagram, I spent eight years working at Instagram, we had a billion users, or billions of users maybe. I don't actually know what they're at now, but it was a lot of people. But the product was pretty simple, and there were a couple ways you could use it. We did think about creators and larger accounts that had special needs, but this is vastly different, where you have someone who's using this to help them come up with the recipe to eat tonight, or brainstorm their next haircut or something like that. And then there's somebody who's transforming the way they're doing farming, helping to automate their farm in Japan or something, right? And it's just the broadest spectrum ever.
Generally what we've seen is that we believe in this idea of a capability overhang, where the vast majority of the people are getting a sliver of the true value that they could out of the bottle. That's not necessarily a bad thing. They're still getting tons of value. But we make sure to focus on creating an extremely simple, streamlined experience for many, many people. And then all of the cutting-edge stuff, we also make sure to give that and make sure that's really accessible for the people that care.
The challenge of course is not overindexing and putting something that people might not understand in front of a billion users. But generally what we've seen is we can experiment early with the people that really care, and that's where we have our desktop app and we have Codex, and now ChatGPT Work is kind of stepping into a much broader audience, which we're really excited about. And eventually what we know, and we think about this a lot in design, is that things will truly become for billions of users when they don't have to think about a switch or a mode or anything like that. And so we know that eventually it has to kind of get distilled into that main experience.
But it is a really fun thing to balance, and we have to think about people that are using this for free every day to people that are paying us hundreds of dollars, and lots of different needs. It's just a matter of balancing, and as a designer it does add a lot of complexity for sure. You really have to have empathy for every different type of user. But it makes it fun for sure.
That is such a mind-blowing point you're making about the spectrum of usage of ChatGPT and Codex. I didn't think about it that way. Has there been a product that has such a wide spectrum, from just "what's this rash on my finger" to building Salesforce, you know?
Yeah. And it makes it really hard. It's so funny, one of the big challenges we're always thinking about is this blank box problem, right? And people talk about this all the time. It's like, is chat really the final frontier? It's a glorified terminal, is some critique that I think these chatbots get. And then it's like, okay, so how do you design something that can kind of shapeshift into anything? Maybe someone is coming here for image generation for fun, or they're trying to generate a PowerPoint. So having this adaptable interface is a really interesting design challenge, and it's sort of uncharted territory because it can really shapeshift into many different tools, and it's incredible at so many different things, and trying to figure out how to present that.
And then we think a lot about a novice user, someone who's brand new to this experience, that we want to help understand how this can add value to their lives or to their professional work or whatever capacity they're using this in. And then we also want to get out of the way of people that use this every single day and use this as a real workhorse tool. And so designing for that spectrum is something we have to think about a lot. But yeah, it's all a world of spectrums, I guess.
There's a comment that Kevin Weil, your former chief product officer, made on the podcast that has really stuck with me, which is, to your point, people always diss this idea of it's just a chat box. Why hasn't it ever changed from that? Even though it has, but still it's kind of this chatbot. And he made this really good point: what is an interface that is very good at spanning a wide spectrum of intelligence? It's actually just talking. If you think about humans, we talk to people of low IQ to very high IQ. We can all just talk to each other, and it's actually a really good interface for packaging something that is increasingly smarter and smarter and smarter.
Yeah. And I think what's been a really fun design space for us is, there's two things we think about. One is how can we help you express your intent better, right? We don't want you to have to think about being a perfect prompt engineer, but how do we even teach you that, oh, this is something that the model can help me with, and then how do we make sure that we best understand what you want. And so we recently shipped some features where if you ask an open-ended question, it can follow up with some follow-up questions that you can just tap through and answer, versus having to type everything out, right?
Or another good example is, once you actually ask it to do something, the chat transcript doesn't need to be purely text, right? We've seen these now evolve so far from what it was when ChatGPT just launched. It's no longer just a pure chatbot. It actually has many features within that, right? So, a good example is what we call writing blocks. You ask it to write you a letter or an email, and it now contains that email in a block that you can directly manipulate, and you can type with it, and you can brainstorm, and you can edit little bits and parts of it. You can copy just that, and that has made that interaction much better. And we're seeing the same thing with image generation and creating special tools for that kind of mode.
And I think that's where the UI is going to start to really continue to evolve. If you come at this and you're a designer, maybe ChatGPT should actually look different or have different affordances. Or you're a data scientist, or you're using this to run your home, your personal life, maybe it should really kind of work or look differently. So understanding the context and giving you different affordances based on that is something that we're thinking a lot about.
In terms of the vision and future of where all this goes, I know this is always evolving, but is there a sense of what the future of ChatGPT and Codex is? I know there's this idea of a super app. You talked about voice, this personal assistant idea. What's a way you think about where this goes long term?
I think that it will start to become more and more adaptive to all of the context of your life. I think proactivity is a big untapped part of the experience that we're really excited to think about. We're starting to see this with ChatGPT Work now, where it actually has pretty good suggestions based on your calendar or your Slack or these different inputs and external context. I think that's going to become more and more important, or kind of shift the way you use it.
I think voice, as you know, we just had a step change in the model that just feels much more natural, and again, I can only imagine how that's going to feel in 2 or 3 years as it just gets better and better. I do love interacting with ChatGPT, especially voice, when I just want to really quickly say something. And then I think its ability to go not just from asking questions but actually doing things for you, that's all going to change.
And hopefully all of this gets simpler as we do it too, right? When we think of the super app, especially for our broad consumer base, we do think that there should be this one universal input that you can ask, that knows when it needs to answer really quickly, or it should be chatty, or whether it should go off and do work for you. And so that's a big excitement, that we can get to how we can continue to simplify the experience such that you shouldn't have to think about how this works. You shouldn't have to think about what model you're on or what speed you're using. That goes back to balancing the different personas and user needs, right? Of course we want to expose that for the right people, but then we want to make sure that we can get a default experience that works great for most people.
But yeah, I think it's going to be really proactive. I do think that we're going to get much richer input through voice and richer visual outputs that are more interactive. I think that you'll start to build more durable workflows with ChatGPT that you can use over and over. Right now, a lot of times it feels like you're coming and starting from scratch each time, and we want to really make this the tool that you can use to help you with whatever task you have.
This all sounds wonderful. I can't wait.
Just kind of zooming out a little bit
And then I want to talk a little bit to designers again, but you're in this epic role right now. I imagine there's a lot of surprises as you entered OpenAI and took on this role. What's something you wish you'd known? What's something you wish you could whisper into Ian of day one and to give you some advice?
I mean, so many things. I think the first one is like recognizing that we're really early, and instead, like, I actually was pretty intimidated coming to OpenAI, if I'm being honest. I remember, okay, interviewing. Oh no, I got to... GPT-4 had kind of recently come out, and what GPT... I was like, what is all this stuff? I was using it and I was using the stuff, but I didn't feel like I had a true understanding of it. And to be honest, I still have so much to learn. So that idea of like, you are early, we're figuring this out, that I think I would want to hear.
And the other thing that I found, and for me this is more of a personal thing, it's less about AI or anything, but I stepped into a really different role. Most of my career I've managed. I definitely managed before, but never at a scale that's growing so fast, and then at a company that is so different, that the role was just... every day was something new. And for a long time I felt like I'm kind of bad at this, or I don't know what I'm doing. And then I went and talked to friends that had led really large design teams or were in a similar transition as me, and they all felt exactly the same, or the problems that I had were not unique.
And so this is, I guess, more personal advice: I think at any stage of your career there's that feeling of impostor syndrome and like, what am I doing? I don't know how to do this. Finding people you can talk to that are going through similar things, you'll find that there's usually a lot of shared challenges and struggles. And having that, whether it's a mentor or just a buddy that is impartial and out of it, was really helpful to me. And finding that community I think has been helpful. So yeah, like that, and then understanding that this is all so new and evolving really quickly.
That is extremely helpful to hear. Elena Verna had this really great post. I don't know if you know her. She's head of growth at Lovable right now. She had this great post called "Stop the AI Confidence Theater," where all these people just keep posting all these things they've done with AI and all the ways it's changed their life, and they're killing it, figured it all out. And really, very few of these things are actually that great and working, and it's just like, stop pretending like you figured it all out.
I think it's much better to be humble about it and be like, wow, yeah, I don't know, this is changing quickly, and I'm just going to embrace it and see where it goes.
Yeah, and that connects to what you said was a core component of people that are doing well right now: people that are curious and just kind of poking around and trying these things, and trying them again because maybe it didn't work before, and now it works.
100%. Yeah. It's like being kind of like... yeah. Every model, every new feature, the people that are poking around at those edges, I've found, are just having such a blast doing it.
And this connects to something else Kevin Weil, former CPO at OpenAI, said on the podcast, which has always stuck with me, which is the model we have today is the worst the model will ever be.
Exactly. I mean, that goes back to like when we... yeah. I mean, a new model comes out what, every week? I don't know, every day. If you count all the labs and everything, it's changing so quickly and everyone gets better. And I always try to remind myself it must be like, I don't know, building or designing in the early days of computers, when componentry was really evolving really quickly and there were these drastic step changes, and you would hit these limitations and all of a sudden those limitations were gone. I think it's the same, right? These things just change, and what was totally broken yesterday now actually just kind of works.
So maybe to start to close out the conversation, say a designer is listening to this and they're in that bucket of overwhelmed, anxious, struggling, stressed, feeling uncertain about their future. What's one suggestion you have for them to help them move out of that state and into a better place?
Just recognizing that it's a totally rational feeling, and it's a great time to just try to explore and see: what if I tried working... what if I tried rethinking the tooling, or what if I just embraced, oh, maybe I should prototype more, let me try. Try using these tools. But again, recognizing that... maybe it helps to hear that I think we're all still so early and it's still changing, and just kind of embracing that. Just have fun with it and see where it goes.
And I think maybe another thing I would say is really try to focus on the output, the outcome, right? At the end of the day, we're not here to build tools, like processes. We're not here to build or token max or something like that. We're here to make great software that people love, or great products, or whatever it is you do. That's the thing that really matters, and if you focus on that, I think you'll hopefully find that these tools can help you do that, but you still have a huge role in that as the person that's kind of steering and directing it.
So, along those lines, I'm going to take us to a couple recurring corners of the podcast. The first corner is AI corner. The question is just, what's one way that you have found AI to be really useful in your job that maybe might inspire someone to use, say, ChatGPT or Codex?
As a designer... I mean, nowadays I spend less time in the pixels, but I really do use it to kind of help me think through broader ideas. Or I'll have an idea at night or something, and then I'll just quickly type it in and have it prototype something, and it'll give me an idea that I can go and talk to the team about, and it helps visualize something. And so that's a part of the design process. It's early ideas, or we're riffing through something, and I've found that ChatGPT, whether it's ChatGPT Work or Codex, can really kind of help me visualize that. That's been something that's been really fun.
And then there's more of the operations side. I have it now checking and making sure that important Slack messages that I need to follow up with are summarized, and giving me context before important meetings, or for recruiting, or for every part of my process. I use this to kind of augment the work that I do.
That makes sense. On the prototyping piece, just to give someone a tip on how to do this, are you opening up your phone and firing off just a Codex thread? What's the mechanic of doing this?
ChatGPT Work has been incredible because it now runs fully in the cloud, right? And so the idea is you can kind of just be anywhere and talk to ChatGPT and have it really spend the time and come back with a really great artifact. So I've found that it's been really fun to just be like, "Hey, we're working on, I don't know, whatever this thing is," whether it's to help me think through it, or, yeah, sometimes I will just have some ideas that I would love to see quickly visualized before I even take it further, and it will just kind of help me play with it and try it and stuff. And so that's been really fun. The fact that it's so accessible, right? I might not even be near a computer, or I can, I don't know, just kind of have that flexibility, I think is really fun.
Awesome. Okay, I'm going to take us to another corner: failure corner.
Yeah.
So the reason I do this: a lot of awesome people like yourself come on the podcast, and all people hear is all the amazing things they've done, all the ways they've been really successful, and they don't hear all the times things didn't work out. And I feel like that's very important for people to hear. It's not always up and to the right and success. What's a story of a time you failed in your career, and what did you learn from that?
I mean, there are so many times. I think, even at OpenAI, I've found that I've had to learn every step of the way, right? When I first joined, I remember not really even fully understanding how I wanted to build a design team, and not understanding how to hire or whatever. I sort of felt like I was kind of failing every day, but just kind of quickly trying to adjust and learn. I don't know, that's maybe kind of a cop-out answer, but really just feeling like every day I'm kind of learning on the go. I feel that my whole career has sort of been that way, and that I have never really planned any one step, and it's sort of quickly learning from your mistakes.
I've shipped plenty of things that did not work out, and I've also shipped things that have worked out, but many times you try something and it's kind of a failure. At Instagram we had flops, and then we'd quickly move on. I worked on IGTV, which was a huge flop, and then we shipped Reels. And that's a good example, because we just kind of baked in some wrong assumptions and we held on to a few constraints, but then we learned from that quickly and we were like, okay, actually let's change that. And then we sort of redid it, and then it became a really successful product.
And I've seen many cases of that, and I think that's helped harden me to the critique and feedback of things, because sometimes you get it wrong, sometimes you try something, and sometimes it is just about having to put it out in the world and work through it and get the feedback and iterate. And I think what's more important is not what you shipped one time, but how you react and how you kind of fix forward and how you continue to iterate and listen to your users.
This is a very common theme in this corner: people's careers are mostly failure along the way.
Yeah, definitely.
And so I think that's helpful for people to hear.
Yeah.
Okay, before we get to a very exciting lightning round, is there anything else you wanted to share that we haven't covered?
If I could distill it, I hope that this is just a really fun time to build stuff. And I hope that we as designers can be really adaptive and focus on making really fun, making really great stuff. And I think great companies are seeking designers and seeking great design, and I've found that to be really true. There are probably going to be a bunch of companies that will be left behind because they haven't shifted, but I think that there's plenty of room for amazing, great design, and that's going to be more true than ever.
I completely agree. I'm very bullish on design.
Yes.
Well, Ian, with that, we've reached our very exciting lightning round. I've got five questions for you. Are you ready?
Okay. Yes.
All right. Here we go. What are two or three books that you find yourself recommending most to other people?
You know, it's funny, I'm not a big book recommender. I do get asked about books for design, and I haven't found that many that have really stuck with me, purely design. Although I will say I do love The Design of Everyday Things. I think that's such a classic. That's the one. It's not a unique recommendation, but it is just kind of a classic of thinking about how we think, how I think about design.
You have to read it if you're in tech or not in tech.
Yeah.
Awesome. Anything else there, or should we keep moving?
We can keep moving. Yeah.
Let's do it. Favorite recent movie or TV show that you have really enjoyed?
Okay, so I have one that actually is not a movie or TV show. I'm in New York. I just saw a Broadway show called Maybe Happy Ending.
Yes. I've seen it. Love it so much.
Oh, you saw it? Okay. Yeah. It was so unique, right? And of course it kind of hit home a little bit because it's about these super intelligent humanoid robots, but the production was incredible. The set design, everything about it I thought was so unique. So if you have the chance to see it, I thought it was really, really good. I also just saw The Invite, which I thought... have you seen it?
Yes. Also love it. So simple but so good.
That's what I loved about it, it's so simple. It actually is like a play, and it's just four people in one kind of beautiful apartment, and the set design is amazing. But the writing, acting, everything about it was incredible, and it's just kind of a wild ride. And I went in knowing nothing about it. I thought it was incredible.
Excellent taste, sir. I love both those so much. And The Invite just came out. It's in theaters right now.
Yeah. Next question. Favorite product you've recently discovered that you really love?
I feel kind of like a typical Silicon Valley person recommending this one, but I did finally switch to an EV after having many cars for a long time, and I got a Rivian. I do respect the design a lot, of just how they've really made it... probably Tesla's similar, but there's a lot of assumptions about how operating a car should work that have gone away. You just walk up to it, it unlocks. You don't have to turn the car on and off. And it's cool to see that evolving. Of course, I love Waymos. If I could own a Waymo right now, that is the most magical experience in the world, and I'm waiting with bated breath for that to be broadly accessible for everybody.
Do you have a favorite life motto you find yourself coming back to in work or in life?
Not necessarily, other than, kind of trying to sum up how I felt about all of this: have fun with it and just try stuff. I've never felt like you have to fit into a specific box, and know that it's perfectly all right to be very fluid about the way you work or the way you think about things, and, I don't know, be flexible and adaptable. But yeah, it's not quite a life motto. More of a vibe of just trying to have fun with it and build great stuff.
I should shift this question to what's just the vibe. What's the vibe that's been helpful to you? Because I love that. It's very a
Sam Altman in terms of just, you can just build stuff.
Exactly. Yeah.
Okay. Last question. Something I love about your career history: if you look at your LinkedIn, you were a designer at Groupon. I love that you were at Groupon and now you're heading design for essentially superintelligence. What's something you learned at Groupon that is maybe helpful these days building superintelligent AI?
I love that. Groupon was a really fun company. Andrew Mason was a really incredible CEO, and he brought such character to the product. It was kind of a brand-led product. We had a comedy team. There was a writer team, but they were all comedians basically, and the product really did not take itself seriously, at least in its early couple years. It didn't take itself seriously, and I think that the brand really stuck out, right? You weren't signing up for Groupon necessarily... I mean, of course the daily deal was great. You were getting like 50% off something, but people loved that email because it was really well written and it was funny and it had character. And I thought that early on the company did a really good job with brand and embracing that side of things.
And I think the other thing, if I were to carry that forward, is I found that every company that I've worked for has such unique DNA, and the way you have to work maybe changes a little bit based on that, right? Groupon was run one way. Then I went to Instagram, which was very design and creativity focused, and Kevin Systrom was really an incredible designer, and he sort of saw himself, or worked, in that way with us, which was really cool, and it just created a different way of working. And then you have Sam Altman, who has a completely different way of operating, and then OpenAI being a research lab. And so learning that companies have these really different DNAs, and understanding how to embrace them and leverage that, or help empower that, or I guess lean into that, has been something that I've learned a lot from moving around and getting the opportunity to work at very, very different companies.
And also just to add to that, it's also important to realize you're not going to change that. You're not going to come in and change that culture, because that's very hard, especially if it's very founder oriented.
I think it's about finding... At OpenAI, yeah, sure, we spent more time at Instagram, we definitely moved slower, and I would say that we changed the product less, and that actually worked really well for most stages of Instagram. But I don't think that would work in the current stage of where we're at with OpenAI and that way of working. Of course, I bring a lot to that. At Instagram we had this motto, "Do the simple thing first," and we always used that and we really embraced that. I've tried to carry that through to the way I think about building product at OpenAI. But that's not the only way of working. Sometimes it's actually not necessarily the right way to think about it. Sometimes we actually do throw some things out that we're still experimenting with. And so I found that I have to be flexible, and not necessarily try to change it, but try to help round it out, maybe.
Ian, what an incredible place you have in the world right now, just being on the inside of such cool technology, such cutting-edge stuff that's happening. And I feel very fortunate that you found time to chat with us and share what you're seeing, where things are heading. Two final questions: Where do you want to point people if you want to plug anything, and how can listeners be useful to you?
Where do I want to point people? I think, please, have fun with the stuff we're making with ChatGPT and Codex, and fire away feedback and thoughts. I'm on Twitter. I love to talk with designers at all stages, or not just designers: engineers, PMs, whatever. That's probably where I'm most reachable and accessible.
And how can you be useful? Again, I love when people show off the way they are working and how that's shifting. That's a great way for us to learn how we can build better products, both internally at OpenAI but also as a design community. Maybe showcasing a little bit more of our process, because it is becoming a little bit more opaque, like, well, how is it working? I think sharing that is such a gift to the community, so doing more of that.
Also, point people to your website, which I have here now. And I think my face is probably changing colors because the site has all these crazy colors. So if you go to iansilber.com, you can see this insane website. That's your personal website, and there's a...
Yeah, that's a deep dive into all of my past work. You can see it there. Yeah.
And the trick is, you'll have to discover the Easter egg where there's a game involved.
That's right.
See if you can set the high score.
Yeah. The high scores are out of control. I don't know what's happening. Someone's found a cheat code.
Ian, thank you so much for being here. Thank you so much for doing this.
Buddy, this was awesome. Super fun.
Same. Bye, everyone. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review, as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at lennyspodcast.com. See you in the next episode.
Article published · Updated