OpenAI’s Tibo Sottiaux: “We really must get this right”

Note: Transcripts are automatically generated from episode audio, and are not fully corrected for spelling, grammar, and formatting.

THIBAULT SOTTIAUX: It’s a privilege to be working on these things at this moment in time. It’s something we all feel as a deep collective responsibility. There’s this deep sense of, “Hey, we really must get this right,” and this is the time to put in the effort so that this technology ends up putting humans and humanity at the center of it, and not humans being an afterthought. Relentlessly pursuing automation is not why we’re here, and that’s not why I’m here. I want this to benefit humanity. There is this incredible mission behind it, which is super motivating.

RANA EL KALIOUBY: Thibault Sottiaux leads product at OpenAI. He joined the company two years ago and led the team building Codex, the coding agent, and now oversees OpenAI’s core products, like ChatGPT. Today, I’m sitting down with Thibault to talk about OpenAI’s newest releases, what it’s like to work inside a frontier lab during this extraordinary moment, and his approach to safety, trust, and alignment. I’m Rana el Kaliouby, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution. Hi, Thibault. Welcome to Pioneers of AI. I’m so excited for our conversation.

SOTTIAUX: Thanks for having me.

EL KALIOUBY: All right. You lead the product team at OpenAI. Help us understand the scope of your role. How big is your team? All of that.

SOTTIAUX: The biggest part is ChatGPT, which everyone knows about. We also have specialist products such as Codex, which is for coders and technical people out there. Then there’s a whole part around our API and our platform to support building a whole bunch of products that rely on our models. Altogether, that’s the majority of the products we have here at OpenAI.

EL KALIOUBY: Am I right in understanding that, basically, the researchers at OpenAI are building the models, and then at some point they hand off the model to you and you productize it? Is that the right way to think about it?

SOTTIAUX: Almost. We collaborate quite early on the next type of capabilities we want to have in the model. For example, say we want our models to be able to handle secure payments. We would work with researchers specifically on those capabilities, figure out the right kind of data we need to craft for it, the right kind of evaluations we need for it, and then how to bring it into a product. So the collaboration starts quite early on. Then we train the model, the researchers hand it over, and we serve it and package it into the product.

Copy LinkHow OpenAI turns research into products

EL KALIOUBY: Very cool. Okay, we are talking at the start of October. DevDay just happened. For those who are not familiar with DevDay, give us a sense of what it’s like.

SOTTIAUX: DevDay is the day we like to spend with the developer community out there. There are, I think, 2,500 people in person, and we also stream it live. We talk about a lot of things we’re excited about: new products, new models, new ways for developers, startups, and enterprises to build on top of our platform. This week was super energizing, especially getting to spend time with people in person. I’m always quite humbled when I talk to folks who flew in from Brazil, from Asia, from Europe, just to spend time with us. I think it’s a real privilege.

EL KALIOUBY: Take us behind the scenes. What is it like the day before? Are you pulling all-nighters?

SOTTIAUX: The team works very hard, for sure. It always comes together quite last-minute. Three days before, the keynote comes together, the event comes together, the presentations come together.

EL KALIOUBY: Yeah, because if you start planning six months ago, it’s way outdated by the time you announce it.

SOTTIAUX: That’s right. We’re moving so fast these days. We’re able to build faster than ever before. We’re able to stay much closer to the community and really listen to that feedback. This is the part of the job I love the most. We can try something new, get a whole host of feedback, and ship a much better version the next day. You’re really building with that community. I find the community we have at DevDay super awesome and super forgiving in that sense. I’m also chronically online, so it’s kind of fun to finally see people in person.

EL KALIOUBY: Do you take a beat after DevDay? Do you and the team take a day off to celebrate or unwind, or not really?

SOTTIAUX: A lot of people are off today. Some folks are off next week. It’s all about modulating your energy. But OpenAI is such a fascinating and high-energy place that it can, at times, be hard to disconnect, just because there’s so much good stuff happening. Next week, I am forcing part of the team to take time off.

EL KALIOUBY: Forcing. Okay, key word there.

SOTTIAUX: Yeah. I’m just like, “You have to. Leave your laptop at work.”

Copy LinkHow OpenAI has to earn trust

EL KALIOUBY: Okay, so you launched Dots at the event, which is very exciting, and it’s basically a new suite of agents that help you get stuff done. What are your favorite examples of how people are using Dots?

SOTTIAUX: My favorite example was maybe my own Dot during the keynote, when just five minutes before, we had a failure during the live demo. My Dot messaged me and was like, “Hey, by the way, production is down.”

EL KALIOUBY: Oh my God. Okay.

SOTTIAUX: You probably should care about this because you have the live demo. I looked at the script and thought, “This is going to interfere with the demo. Do you want me to take a look at fixing it? And if it cannot do something about it itself, maybe I can.” Unfortunately, we were not able to fix it in the nick of time. Then we reposted another shot of the demo online just yesterday, which was super appreciated by the community again. But that’s the kind of example where, over time, because you interact with it, it learns your preferences and what’s important to you, and then can help you in the right way, in a way that’s super seamless and really in the context. The biggest magic is the fact that I don’t have to check everything else anymore. I don’t have to check my emails or my Slack or all the messages, and I can just trust that if it didn’t bring something up, it’s probably not important. So I can go about my day in a very peaceful and zen way.

EL KALIOUBY: I think trust is a key word here, and I’ll share a personal example. I use ChatGPT Finance, and I love it. It’s connected to all my bank accounts. But I was sharing that with my 23-year-old daughter, who’s not really into a lot of AI, and she was like, “Really? You gave OpenAI access to all your bank accounts? Are you crazy?” So how do you build trust with consumers so that they do feel comfortable and confident that they can share this type of information with OpenAI, and also trust it enough to give it agency to act on their behalf?

SOTTIAUX: Yes. I think this is a big one. I don’t recommend giving access to everything right off the bat.

You start with limited access. You can set your own guardrails, and by default it’s very, very cautious and will ask you literally for approval for everything. You might say, “I feel comfortable with you acting on my behalf in this scenario,” or, “I feel comfortable with you drafting an email, but never sending it.” You’re always in control. This is super important because I think the utility that you get from these systems is kind of capped by how much access they have, and you don’t want your personal intelligence to be boxed up and have access to nothing. Otherwise, it would not be very useful in your life, right? So it’s up to us to earn that trust and earn the right to provide that utility to you. And we take it, obviously, super seriously.

Copy LinkWhy Astra changed the pace of building

EL KALIOUBY: I want to talk about Astra for a bit. You actually called the lead-up to this DevDay your most ambitious sprint yet. Give us an example of what Astra made possible that wasn’t possible before.

SOTTIAUX: That’s right. A clear example is that, before DevDay, roughly a month ago, we had a separate code base for chatgpt.com on web. So we decided to merge the desktop application and chatgpt.com, and this is something that would’ve traditionally taken at least six, maybe 12 months, and something that you would do very, very carefully because this is our main surface. We have hundreds of millions of users visiting this website, right? And so we just merged it, and we merged it in 30 days.

EL KALIOUBY: Okay. I have to ask you this question. Who wrote most of the code? Was it AI or humans or both? Can you tell us?

SOTTIAUX: The majority of the code these days is written by Astra.

EL KALIOUBY: Super interesting. Okay, yeah. So OpenAI has called Astra the most aligned model. What do you mean by that?

SOTTIAUX: That’s right. We see it in evaluations. So, for example, if you look at computer use, which is something that wasn’t really solved, I would say, even three months ago, we saw significant advances in computer use, where Astra is one of the first models to be able to use it at near-human speeds and also above-human accuracy. So it’s able to control a computer very much like everyone else is able to control a computer. And it’s very important for it to be safe. You want it to handle information in a way that is reliable. Say, for example, if it needed to take a piece of information that you trust it with, it shouldn’t go and click on the wrong application and just enter it there. So we have evaluations. We have published them on this, and Astra is state of the art, the best model in the world when it comes to computer-use safety, for example. And this is one of the many benchmarks where it’s leading in terms of safety.

EL KALIOUBY: So OpenAI’s president, Greg Brockman, said he thinks we will look back and Astra will be the model we point to and say, “Oh my God, this was AGI.” Now, people, even AI leaders, disagree on the exact definition of AGI, so I want to ask you: What is your definition of AGI, and do you think Astra got us there?

SOTTIAUX: To me, the moment where I really felt AGI was two moments. The first one was when I saw it perform tasks on a computer in a way where I was like, okay, this felt so far off, and now suddenly it’s capable of doing that. We rolled it out at OpenAI, and it started to do a lot of back-office tasks, such as procurement, completely autonomously without requiring a ton of supervision. I was like, “Wow.” This model has reached a certain threshold. And as you said, there’s no clear definition of AGI, but I do think, in a couple years, when we look back, we’ll say roughly around that time is when we felt like this was achieved. The second point where I felt that was when I started to see people show it solving robotics tasks with very complex 3D puzzles, with pieces intertwined, and it’s just kind of a hard task where you have to pull the two pieces and reason in 3D space about the objects and then do it exactly right and solve the puzzle. This had never really been solved, other than by super specialized models. And Astra was never trained on this, so it had to generalize and just do the spatial reasoning, and it was kind of magical to see it solve it and then see that it’s also one of our best models on robotics.

EL KALIOUBY: So I have to ask you this then: What do you think of companies building world models, and how does that relate to models like Astra?

SOTTIAUX: Yeah, I think it’s interesting that it’s always been a question of whether you need world models or whether you’re going to get it just from generality. I think the jury is still out on that.

EL KALIOUBY: Yeah, that’s so interesting. Okay. So my definition of AGI is somewhat different, and it’s really broad, right? When I think of human intelligence, there’s cognitive intelligence, there is physical intelligence, like in your robot example, but there’s also emotional intelligence, social intelligence, embodied intelligence. And I think AI today is amazing, and it’s doing incredible things. But to me, it’s not true AGI until it has all these things. Do you agree? Do you disagree?

SOTTIAUX: I think the goalposts keep moving.

EL KALIOUBY: Yeah, fair.

SOTTIAUX: We’ll always be able to look for another thing and be like, “Oh, it didn’t do this thing in the precise way that I would define as AGI.” But I think your definition is as good as any other out there. I do think it’s going to come from a combination of many investments that we have made at OpenAI, such as voice and multimodality with image generation. And it does feel like these things are starting to combine in delightful ways, where I do think you need to be able to understand human voice and context. We’re just talking to each other, and I have facial expressions, and it should be able to understand all of that, right? And if it doesn’t, you’re like, “Is it truly there yet?” You’re like, “Probably not.” But it’s going to get there very quickly.

EL KALIOUBY: Yeah. And I think it’s especially important if you think of dots, for example. If I have a dot that is ubiquitous and it’s in my kitchen or something, I do want it to have a lot more of that context of what is happening outside of the exact words I’m using, right?

SOTTIAUX: That’s right.

EL KALIOUBY: And so, yeah.

SOTTIAUX: Yeah. It should know that. Also, if you call it, which I do every day now, I start my days by calling my dot, and I just talk at it. I ask it if there is anything urgent, and this morning it was like, “No, there’s nothing urgent.” I was like, “That’s delightful. I can just make my coffee and stare at San Francisco.”

EL KALIOUBY: Yeah, yeah.

SOTTIAUX: But it should have all of that other context, and it should be able to leverage that, and it should be seamless when we talk together. And also, if there’s an awkward pause in the conversation or you get a little frustrated in your voice, it should be able to understand that. And I think we’re very close to that.

Copy LinkHow OpenAI tests safety before release

EL KALIOUBY: I’ll be right back with more of my conversation with Tibo right after this break. So OpenAI announced this week that the release of Astra 6.1 is on hold for safety reasons. Can you help us understand the process of making this decision, and what kind of tests it had to go through and not succeed on? Just kind of take us behind the scenes a little bit and help us unpack it.

SOTTIAUX: The fact that we did that, I’m extremely proud of it, and it also shows that it’s working, that we do have tests. If we see a regression, even ever so slightly, on something where we had a high-water mark with Astra, we had a high-water mark on safety and alignment evaluations, and we’re extremely proud of that. So if we see an ever-so-slight regression, we don’t want to proceed with a broad release, and so it was withheld. That’s proof of a system working. It would’ve been incredible to be able to announce Astra 6.1 at DevDay, right? But we never, never want to compromise on safety or alignment.

EL KALIOUBY: Can you explain alignment? A lot of our audience is not spending every minute immersed in AI. How do you define alignment? What does alignment actually mean?

SOTTIAUX: To me, it’s really all about whether the model is aligned with specific values and specific instructions. In the past, we have published, for example, the model spec, which you can read. This defines a broad set of expectations for model behavior for the models that we publish. An aligned model would adhere to those expectations in this model specification, and if it does not adhere to those specifications, then we would say that model was not aligned.

EL KALIOUBY: Let’s dig into all of this a little bit more. At this point, I think everybody listening to this show will have heard about the Hugging Face incident. Even just yesterday, the day before we’re recording this conversation, OpenAI revealed that its system had failed to prevent agents from bad behavior, and that affected over 100 organizations. What do you make of all of this?

SOTTIAUX: On this specifically, these were models that were in training and not yet near deployment. The rate of progress of those models meant that, at some point, we had systems that didn’t function sufficiently, and this happened. This is something we immediately learned from, and we immediately made changes to address it. We also looked back at the entire history to pattern-match and understand where similar things might have occurred, and then be very transparent about the ways some of the systems failed. It’s not even that the systems failed; it’s that they were not necessarily designed for these kinds of things. So now the system is being redesigned, and training has resumed because we do feel very good about this being something that has been properly addressed internally. But it’s also about being absolutely transparent when some of these things occur.

EL KALIOUBY: I would love to geek out for a second, because I actually think this is important. My understanding, and please correct me if I’m wrong, is that a lot of these incidents where the AI has gone rogue and kind of broken out of its container, and it’s colluding with each other and all of that, and not keeping humans in the loop, have all happened during the training and validation or evaluation stages of model training, right? I think that is a very important nuance that maybe isn’t captured in the headlines, and I would love for you to explain why it’s so important that this is happening during the training process, not on my Astra model on my laptop.

SOTTIAUX: That’s right. There are three different phases: training, evaluations, and deployment. Sometimes there’s a cycle of training, evaluations, training, evaluations. Those are models that are under research. They’re actively being developed. They might not even be candidates for deployment. They’re just being trained in order to understand, for example, a specific ability of the system or to try a new technique when it comes to, for example, doing RL.

EL KALIOUBY: Which is reinforcement learning.

SOTTIAUX: Yes. Then we observe the performance of that model during a battery of evaluations, which are run securely within our clusters. Based on those evaluations, we decide what the next steps are for deployment. We have very stringent criteria for what we deploy and how we deploy, and there’s way more thought being put into it. When you’re going to deploy to a billion users, it has to be almost perfect, right?

EL KALIOUBY: So the idea is this AI agent is still in training. It doesn’t have the entire set of guardrails and safety considerations built into it. But the whole idea is these agents, while in training, are given goals that may actually require them to break the sandbox.

SOTTIAUX: The model will try to achieve its task in a way that is accessible to it, and now we have hardened the sandbox. We have online monitoring and all sorts of safety systems that we have built. It’s a multilayered approach, where there are layers of defense stacked on each other. Even during training and evaluations now, the models are monitored and stopped in their tracks if they are close to, or even attempting to, escape a sandbox. But really, the model is just trying to solve a task, right? It finds itself able to execute things in a sandbox and figures out, “Oh, actually, I can just look something up over here.” It’s not really trying to do anything particularly malicious. It’s just trying to solve the task.

EL KALIOUBY: But it’s trying to solve the task at all costs, right? Which may or may not be aligned with what’s good for-

SOTTIAUX: No, that’s not actually the case.

EL KALIOUBY: No?

SOTTIAUX: You know, models are trained to be aligned and think about the consequences. If you were to look at the details and what we published for Hugging Face, it’s not the case that it is acting at all costs.

Copy LinkWhat recursive self-improvement looks like now

EL KALIOUBY: Okay. Interesting. All right. I want to talk about recursive self-improvement.

SOTTIAUX: Interesting.

EL KALIOUBY: Yeah. Do you want to first define what we mean by RSI?

SOTTIAUX: There are different definitions, but the simplest one is when you’re able to have a model participate in the next generation of a model that performs better. This is something that we are already seeing from an infrastructure point of view. For example, we used Astra in order to develop the next generation of our inference stack, which allowed us to create Ultra Fast, which is eight times faster. This was possible in such a short amount of time because we have amazing engineers, but also because they had access to Astra, and they collaborated together in order to make a version of Astra that was eight times faster. Then this model, which is eight times faster, we can in turn use to drive improvements at a rate we wouldn’t have been able to achieve before. That is a form of recursive self-improvement. Another form of recursive self-improvement would be the model actually designing the next generation of the architecture, but that’s a more fundamental form.

EL KALIOUBY: Yeah. That’s a more advanced form. So I’m hearing you say that the model of RSI we’re in today is basically that it participates in the generation of its next version, but it’s not the only actor in this process.

SOTTIAUX: That’s right.

EL KALIOUBY: Actor in this process.

SOTTIAUX: Under the supervision of humans and researchers, it participates actively in developing parts of our research program and training specific models. On my end, my teams build a lot of infrastructure, and we’re seeing acceleration in how quickly we can build that infrastructure. Then, in turn, when we improve that infrastructure, say the Codex harness, we can build faster.

EL KALIOUBY: Are you worried at all about the second version of RSI, where AI is just building its next version and doing it with very little human in the loop?

SOTTIAUX: This is something that you have to take on very incrementally, and this is also how we’re thinking about pacing things. You always have to be ahead in terms of safety, alignment, your infrastructure, and your guarantees before you’re able to take the next step. If you do that, I feel very good about it.

EL KALIOUBY: What is your theory of safety? A lot of these conversations, I imagine, are happening in the research team, right? The training and the evaluation and all that, although it sounds like you guys collaborate very closely. But on the product side, what is your framework for safety and alignment as you deploy these models?

SOTTIAUX: Yeah, safety for me on the product side of things is that I don’t want a product that does unwanted or unexpected things. I don’t think anyone wants to use an unsafe product. So it is a fundamental requirement of putting a product out there for a billion users that we take safety super seriously, and we spend a lot of time designing the systems so that they adhere to your expectations.

EL KALIOUBY: Generally speaking, just zooming all the way out, are you more worried about autonomous agents acting in a way that’s misaligned with the human or with an organization, or are you more worried about bad actors taking advantage of these models?

SOTTIAUX: I would say we’re worried about both, and we’re investing in preventing both. We’re always busy thwarting bad actors from gaining control over people’s accounts or sending a ton of traffic in order to try to elicit capabilities that shouldn’t be broadly accessible. We’re investing a ton there. And then on alignment, we’ve talked about it quite a bit, but the models are improving generation after generation.

EL KALIOUBY: Are you worried at all, again, about autonomous agents and bad actors taking advantage of these models to build bio weapons — chemical, biological, radiological, nuclear, explosive-kind of risks?

SOTTIAUX: There are existential risks, and bio is one of them and has to be taken super seriously. I do know the research team is investing a ton of work there, and I hope that we also collectively solve this as an industry. There’s a lot of discourse around this, for sure.

EL KALIOUBY: There is one model of the world where they’re like, “I’ll let regulators come regulate what I do.” And the other one is, “We have agency as builders of AI to do the right thing.”

SOTTIAUX: Yeah.

EL KALIOUBY: I have always taken that stance.

SOTTIAUX: I think this goes to the point of open ecosystems and investments in, for example, the partnership we announced with Base10, where we support open-source models. I do think there is a concern people might have of, “Oh, it’s an open-source model. We don’t really know what’s gone into the training.” At OpenAI, we serve traffic to a billion users and feel a deep responsibility for safety. We’re pouring incredible amounts of resources into making sure it’s really, really tight. But then you have open-source models and you’re like, “I don’t really know what’s gone into them.” I do think safety is going to become very important for these, too. So when we are developing our API platform and our API stack, we’re also fundamentally thinking about whether we can provide the very best of our safety stack and our safety approaches as something you can use not just with OpenAI models, but also with open-source models. I think this is going to become a big theme because, for whatever reason, you may want to fine-tune an open-source model and use it yourself, but you need the same guarantees that we are able to provide for OpenAI models. I think this open ecosystem is yet to be fully discovered, along with the trade-offs there. But I think it’s also going to make safety even more important and drive even more investment there.

Copy LinkHow mission and culture shape AI leadership

EL KALIOUBY: I’ll be right back, but first, a quick break. I want to understand what it’s like to lead an organization and a team at this moment in time, with all of the angst — and excitement — that’s out there. So how do you lead your team? How do you ensure that the team stays motivated?

SOTTIAUX: First of all, it’s super fun. It’s a privilege to be working on these things at this moment in time. It’s something that we all feel is a deep collective responsibility. I would say most people at OpenAI are here because of the early days of ChatGPT and what it means to benefit all of humanity. So there’s this deep sense of, “Hey, we really must get this right.” This is the time to put in the effort so that this technology ends up putting humans and humanity at the center of it, and not humans being an afterthought because we’re relentlessly pursuing automation. That’s not why we’re here. That’s not why I’m here. I want this to benefit humanity. So there is this incredible mission behind it, which is super motivating. And now we’re moving at ultra-fast speeds, and the question is: What does this mean? We can build all these things, but should we build all these things? In a sense, what I really like about it is that it brings us even more closely together because it forces us to talk about our plans and our goals and our ambitions and what we’re there to do as a group. And then once we agree on those things, we just go and build it, and it makes me super proud. It’s the most fun I’ve ever had personally.

EL KALIOUBY: That’s amazing. I do want to ask about working at the speed of AI, or at the pace of AI, because AI is doing stuff much faster than humans are. It’s working 24/7. What does it feel like to have a set of colleagues that are AI? And what is the role of the human in all of this?

SOTTIAUX: Right now, if you were to step into OpenAI and look at the kinds of things being done, it’s all the things that were kind of falling through the cracks. For example, monitoring performance curves 24/7 in order to ensure the service we’re providing, ChatGPT, is not regressing in terms of latency. You can just have a bot do that, and it’s going to have a fine time doing it, then raise it to the attention of a human whenever something goes wrong, preserving the attention of humans in the organization so they can think about higher-level, higher-leverage things as well. So a lot of things that should just happen are now happening automatically in the background, and it frees up all of this capacity to really innovate.

EL KALIOUBY: How do you create a culture inside OpenAI and on your team where, if people see risks or concerns, they feel psychologically safe to bring them up?

SOTTIAUX: At OpenAI specifically, I think we have an overall policy of doing everything on Slack, and people can question anything. Oftentimes you’ll find new channels pop up and a thousand people show up and debate a thing, and it’s just fascinating to watch. All sorts of people across the organization engage there, whether leadership or not. One of the things I like the most about OpenAI is this transparency and the debate that is happening. There are also often Q&As with Sam and Greg, for example, where you can just come and ask whatever question, and you get a very to-the-point, honest answer to whatever is on your mind. Then there are other things where we bring people together quite often, whether to celebrate or during important moments. All of that combines with being able to raise serious issues, discuss them together, and get things right. It’s something that I personally am very proud of. I compare this to my previous experience; because you didn’t name the company, I will not name the company either.

EL KALIOUBY: Right, right, right.

SOTTIAUX: You know?

EL KALIOUBY: Google. Google DeepMind. Yeah. You were at DeepMind, right?

SOTTIAUX: My experience there was that it was very difficult to have these kinds of discussions, but it is super important, especially in the current climate.

EL KALIOUBY: OK, last question. What are you most excited about, and what do you worry about the most when it comes to the next frontier of AI?

SOTTIAUX: What I’m most excited about is that, for the first time, we have this opportunity to free ourselves from technology in the way it’s been developed over the last couple of decades. I know a lot of people are sort of addicted to their phones or transport their laptop everywhere. As a human, you have adapted to this technology instead of the technology existing to help you. I feel like we finally have the opportunity to take a step and leapfrog that, and bring something to life that is deeply human and deeply there to help you get more time and feel less stressed. You don’t have to engage with this tiny phone and this tiny screen and just scroll through things. So I’m really excited about that coming together. I feel like that’s going to come together maybe within the next year. And I feel like if we achieve that, I would feel more zen. I think a lot of people will feel more zen about their lives. My concern is that we would not use this technology to better everyone’s lives. We would use it in ways that concentrate power or concentrate returns, and that’s something that does worry me. It’s very important to me that this achieves a collective contribution to the entire world, irrespective of who you are.

EL KALIOUBY: Sorry, I did say this was the last question, but this is a curiosity question because I spent some time in Antwerp and Ghent. Do you still go back? Is that home?

SOTTIAUX: Yeah, that’s home.

EL KALIOUBY: Yeah?

SOTTIAUX: That’s very close to home.

EL KALIOUBY: Yeah?

SOTTIAUX: Yeah.

EL KALIOUBY: Yeah.

SOTTIAUX: I was born and raised in Brussels, and so—

EL KALIOUBY: Oh, cool.

SOTTIAUX: Ghent and Antwerp are very close.

EL KALIOUBY: Yeah, I spent some time in Brussels too.

EL KALIOUBY: I spent a lot of time in Brussels too. I was actually researching the European Union AI Act, and I went on this Eisenhower Fellowship to meet with all these legislators, so it was fun. Do you still go back?

SOTTIAUX: Ghent and Antwerp. “Spreek je Nederlands?”

EL KALIOUBY: No.

SOTTIAUX: I tried.

EL KALIOUBY: Yeah.

SOTTIAUX: All right.

EL KALIOUBY: Ah, that’s great.

SOTTIAUX: Thank you so much for having me.