Event Replay: The AI Pit Crew: NASCAR’s Legacy of Innovation Meets the Age of AI

[00:00:00] Speaker 1: Hey, everyone. How're you doing? Before we get started, I want to do a few shout outs. I want to say hello to Joyce Russell. She's a researcher at OpenAI. We love to see our teammates in the forum, and we have more of the NASCAR team here today too. Patrick Carroll, Andrew Odom, Julie Gross. Hello, everyone. Welcome. Welcome to our community. So I'm Natalie Cone, your OpenAI forum community architect, and I'm actually joined today from our OpenAI office in San Francisco, which is so cool because I usually produce these in my home office in Austin, Texas. But I'm here with my team, and this is going to be so fun. We're coming into this conversation with a little bit of race day energy. If you think about NASCAR, it's not just about speed. It's about precision, teamwork, and constant innovation. For more than 80 years, the sport has evolved by embracing new technology while never losing sight of what matters most: safety, performance, and the people behind the wheel and behind the scenes. Today, we're exploring what happens when that same mindset meets AI. What does AI actually look like inside sports, where decisions happen in milliseconds? How do you introduce powerful new tools without losing trust, control, or the human instinct that makes teams great? And maybe most importantly, how do you make this technology useful, not just for engineers, but for everyone across an organization? To help us answer that, I'm joined by two people who are living this transformation every day. Derek Thomas leads analytics and racing operations at NASCAR, working across everything from competition data to fan insights, helping teams understand what makes a sport exciting and how to make smarter decisions in real time. And Rich Bowman leads AI at NASCAR, focusing on how to actually bring these tools into the flow of work, from governance and training to scaling adoption across the organization. So that AI isn't just an experiment, but something people trust and use every day. What I love about this conversation is that it's not theoretical. This is about real workflows, real teams, and real change happening inside a legacy organization that knows how to evolve. So as you listen, I'd invite you to think about a few things. First, what does it take to actually empower people to use these tools well, not just give them access? And how do small experiments turn into something that scales across an entire organization? So whether you're deep in AI or just starting to explore it, NASCAR turns out to be a pretty great place to learn. To get us started, I'd love to welcome Derek Thomas to the OpenAI Forum stage. Derek, welcome.

[00:02:52] Speaker 2: Thank you so much, Natalie. Appreciate you having me.

[00:02:54] Speaker 1: Absolute pleasure to have you. And you're joining us from North Carolina?

[00:02:59] Speaker 2: Yes, ma'am. Here in North Carolina.

[00:03:01] Speaker 1: Awesome. So happy to have you, Derek. Well, let's just jump right in. So when people hear AI and NASCAR, they may first think of autonomous vehicles, but that is not what we're talking about today. NASCAR is already one of the most data-rich fast decision environments in sports. So my first question for you is, what kind of data does a team actually have at its fingertips? And how can AI help turn that information into better decisions on race day?

[00:03:31] Speaker 2: Wow, there's tons of data. Just looking at Pit Road on race day, all the screens, all the laptops, lots of data coming in for decision-making. Information like speed, RPMs of the engine, certainly how much time drivers are on the throttle or on the braking. A lot of it goes into decision-making, and importantly lap times. We think about going around the lap; it depends on what track we're talking about, but it could be 30 seconds or so for an initial lap or two. But over the course of time, you're several, like 10, 20, 40 laps; that grip that's so important to drivers, that helps them go faster can wear off. A driver told me once at a testing session, think about it like any of us on a highway. Whenever there's a high downpour of rain, you kinda let off the gas pedal, right? Because you wanna make sure you don't lose grip. Same thing with the drivers and grip the tires. So they wanna make sure that they have the ability to go as fast as possible. But as it wears off over time, naturally at the highest levels of performance, there's a choice that needs to be made to come in for fresh tires. And when does that happen? Really depends on how other drivers are doing and the information that's available to them for that decision.

[00:04:40] Speaker 1: Well, thank you so much, Derek. And in racing, performance and discipline tend to go hand in hand. So how does that philosophy show up when your team is working with advanced analytics or AI?

[00:04:53] Speaker 2: So every Tuesday we get together and have a post-race debrief. We talk about the inspection process, we talk about anything...

[00:04:58] Speaker 1: The inspection process, we talk about anything that happened during the race, but we also talk about the metrics. The people who are in the room talking about this are the engineers who helped design the car. They're people who are innovating the car. They're people who are also officiating on race day with the inspection process and so forth. So having metrics provides them insight. It's kind of like looking under a microscope at the race afterwards. So we'll look at these different data sets. But importantly, when it comes to using or leveraging AI, remember the comments I told you about from fans. Not only are they rating the scale, zero to 10 for the race, but they're also giving us feedback. That unstructured data can be so time-consuming to analyze and also making sure that what we present to our leadership is representative of what fans are saying. So when we can bring those together, that gives us a really good idea of how to assess the race. I'll also be careful to talk about how we get started with that analysis because it can be very... you need to be very careful when it comes to leveraging the tool. It's a lot like using a bolt in a nut. You wanna get started very carefully, right? You wanna make sure you don't strip the threads of that bolt or the nut. So once you get started with the appropriate prompt, then you can start ratcheting down and get more efficient. And that's how we really leverage this information, especially the unstructured data to get a good representative look at the information.

[00:06:24] Speaker 2: That really makes sense, Derek. And it also reminds me of, we have some good friends at the San Antonio Spurs, NBA. And similar to NASCAR, they've been really successful in the widespread adoption of AI tools, specifically open AI. And they talk a lot about laying the foundation and approaching it in phases. So it sounds like... I'm gonna have to introduce you to them. It sounds like you guys are actually creating the playbook for making AI work in the realm of sports. Like I keep, I'm hearing these through lines which is really exciting.

[00:07:02] Speaker 1: Well, we're really still learning. I will tell you when we started this role, I started this role about four years ago, we had a, what we thought was a pretty good idea of what were some KPIs that we would wanna look at. What is some information from the race cars that will be most indicative of the quality of the race? For example, how close is the lead car in the second place car throughout the race? At the end of the race, we call it the margin of victory, right? But we wanna know what's that gap throughout the entire race. Sounds like a good KPI, right? Another one would be how much a certain driver led the race. Was there a dominant driver or do we have a mix of different drivers? And so what we did, we took data from the fan feedback and data from the race cars after three years. And we said, basically we put it into the tool and asked it, what are the, are these the right indicators? Is this what we should be looking at? And the tool came back and said, yes NASCAR, those KPIs that you're looking at, the early three that we were looking at are good, but there are two others that you should be keeping your eye on too. And this is really where it got interesting because while we were looking over here in terms of the gap between the lead car and the trailing car and the amount of leading laps led, the tool is also telling us, not only is it important to have a close battle for the lead that time, but also you have to have the actual passes. So now we're also evaluating the amount of passing between two different drivers at the very front. It's good to have passing for 20th and 15th and maybe 30th, but the real important indicator is up at the very front. And that's what the data has told us by taking a closer look at it. I'll also point out to you that we continue to look at those now five KPIs and it's given us good credibility, I think, as we look at the fan feedback with it to go along.

[00:09:00] Speaker 2: Yeah, and Derek, how long have you been doing this? Cause I'm curious what your life looked like before adopting AI into your workflows and your data analytics practice? And now?

[00:09:01] Speaker 1: That's a great question. So we started our partnership, I think it was November of 2024, and I was very careful not to use any AI tools. I knew there was some sensitive information I had. I wanna make sure it wasn't getting released. I wanna make sure it was secure. So I didn't even play with AI. But whenever I got the green flag to do so and took the training and learned how to put the instructions in there and really figure out the right prompt, that's really where things got cool. And so to answer your question, it's really been a little over a year that I've done it, but it's come quickly just because I think that's natural instinct, that natural curiosity to learn more came out. And as you use your prompts and you modify the prompts, I think that's really where it got to be interesting. So it's a long way of answering your question, but a little over a year.

[00:09:43] Speaker 2: No, that's really fun. So, I mean, that truly makes NASCAR pioneers. I mean, you guys are absolutely ahead of the curve and early adopters. What made you feel like... what was the instinct to, in a way...

[00:09:56] Speaker 1: Instinct to, in a way, take a risk on a frontier technology? So, it really is an efficiency play. It was taking me hours, literally hours, Natalie, when I would go through these spam comments. I don't have a whole lot of time between a checkered flag flying on a Sunday evening and having a Tuesday morning post race debrief. There's not a whole lot of time to write, to put together analysis. So, it was taking a lot of time. And just think about the human. I mean, as much as I was trying to do a good job, I mean, there's always the human factor that could be skewed one way or the other. So, not those amount of time, but seeing that this was gonna be a way for us to come through all of the data quickly and get a good representative look at what fans were telling us and sorting it by what were the most common things or common themes that people were saying. That was changing. I'm now down to minutes on a Monday afternoon, and so I'm prepared, certainly when I get on on Tuesday, to talk about that.

[00:10:55] Speaker 2: Yes, that makes so much sense. My colleague, Caitlin and I, same. In the Open AI Forum, we went from having 200 members to 70,000, and we wanna make sure that we're responding to them and not just responding physically with emails, but responding by curating programming that resonates and that is meaningful to them, and ChatGPT helps us so much with that. So, I really get that piece.

[00:11:18] Speaker 1: We definitely see it, and we certainly have focus on our leadership; once all transparency, we even have our good partners at Goodyear come into our meetings. They see the same exact dashboards that we're putting together so that they can help us. I mean, we can't do this alone. It really comes down to partnerships, and so when we do see certain tracks or track types that haven't, let's just say, performed as exciting as we had hoped they would, and we're seeing a pattern of that, we can go to our friends at Goodyear and see what we can do to maybe change the tire compound, and they've brought different tire packages, and so we're exploring that, and how do we have good tire wear so that the drivers have to manage their resources throughout a green flag run? Another example of that is increasing the horsepower. So we found that in similar tracks, we needed to make sure that maybe there's a way that we could give the drivers more power in the cars, which we require them to maybe manage that resource because more power also means maybe chewing up the tires a little bit more. Maybe it means a little bit different in the braking zones when they're going into a turn, but the other day, it was working with engine builders for those teams that run Chevrolets and Fords and Toyotas in our sport to get an idea. Is this gonna be a hardship or is this gonna be a good thing? And so far, we're seeing it's a good thing.

[00:12:29] Speaker 2: That's really exciting. Can you tell me a little bit about how you've been leveraging the tool in more specific use cases involving fan engagement? Because that fascinates me. And I think fan engagement, it has a lot to do with what we do in community building as well. I always turn to sports and really trusted brands in sports that are able to connect with human beings. And I always say, when you pass someone, I moved from Texas to the Bay Area back in the day, and when I would pass someone with a San Antonio Spurs hat and I'm really far away from home, you feel this sense of camaraderie. And when you pass someone with a NASCAR T-shirt, and that's also really your jam, you feel this sense of we're connected even though we don't know each other. So what have you learned about fan engagement or what have you learned from leveraging AI and your fan engagement analytics and behavior techniques?

[00:13:23] Speaker 1: It's just cool when fans realize that we really are listening to them. And so we may not be able to provide feedback directly to each and every fan's comments, we are listening. And having the confidence in the data, which we do have from whether it's from the race cars themselves or from the fan feedback and making sure that we're getting a good representative and having the confidence to know that we're making the decisions that are right for the sport that are gonna move the needle and make it more exciting. I think that's really the key here. It's really listening to the fans and showing them that we're listening. So that could come in the form of a variety of tracks, the variety of the schedule that we have. But also when we do have exciting races, make sure we don't do anything to mess those up. And when we have races that may not turn out as exciting as we had hoped, let's see what we can do about it. We don't want to jump to conclusions. So we do need to make sure that we have a good sample size of races and then act accordingly with our partners.

[00:14:23] Speaker 2: Awesome. Thank you so much, Derek. So last question before we bring on your colleague, can you walk us through another real-world example that we haven't touched on, we haven't gone deep yet, and how your team is using AI to turn race data into insights and what that process looked like then versus now? Just we really want to ground this talk in real-life experience for the audience.

[00:14:48] Speaker 1: It comes down to efficiently going through the data. We have always received fan feedback for really going back to 2009.

[00:14:54] Speaker 1: Going back to 2009, which was several years. Every single race, we would measure the fan perceptions of the racing. The ability now to combine that with a second data set, I'm speaking specifically about the race cars themselves, has given us a better idea of the quality of the racing. So let's say you're making French fries, you wanna make sure that you deliver them with the right saltiness at the right temperature, that they're not overcooked, not only at the same restaurant, but then maybe across the country. That consistency is so important. We're looking for that same consistency in the high quality of racing, regardless of which track we go to. And yes, every track is different. It may be the same track length, maybe one track is a mile and a half and another track is a mile and a half, but they each have their own characteristics. How do we deliver the best racing on the planet? And being able to use this data, combine these two different data sets together, has enabled us to confidently make the decisions that are best for the sport, to make it most exciting for the fans and really leverage that. That's really exciting. And as I told you as we were planning this talk, Derek, I'm fascinated with just amazing athletic abilities, but also professional sports. What you're saying reminds me of the Lance Armstrong documentary about how his last coach, he was really just tuning in and tuning all of the little details, every tiny little thing about the data, analyzing the track. And it was in the aggregate where they built the winning team and the winning racer, and now you have this incredible tool that helps you go deep on every single one of those little nuanced moments of data. And knowing where to focus your resources, right? Again, you don't have to change the whole spectrum. It's just, there are certain areas that probably need more attention than others in this and having the data, the ability to analyze it efficiently, quickly, and deliver it to the decision makers, the people who can make the improvements is where it really matters. So efficiency and focusing on the right resources, focusing the resources.

[00:17:08] Speaker 2: Awesome, totally. Well, thank you so much for sharing your experience, Derek. This was a lovely, fun time, and I hope it's just the beginning. I'm gonna send you backstage for a few moments now and invite your colleague, Rich, to the stage, but we'll bring you back for the audience Q&A in just a few moments. Thank you for joining us, Derek. We'll see you in a few moments. Thank you. Please help me welcome to the stage Rich Bowman.

[00:17:31] Speaker 3: Hey, Rich.

[00:17:33] Speaker 2: Hey, Natalie. How you doing?

[00:17:34] Speaker 3: I'm doing so excellent. I'm having a really fun time in the studio. As I mentioned before, usually I'm alone in my office conducting these interviews, but I have my team here with me, so this is gonna be so fun. And Rich, where are you joining us from?

[00:17:49] Speaker 2: Also North Carolina. We're here in Charlotte.

[00:17:52] Speaker 3: Oh, awesome. Charlotte is a beautiful city. So let's dig right in. Rich, first of all, congratulations on your promotion. You've been deep in the day-to-day of AI operations, and now you're helping shape how it scales across NASCAR, so congratulations. Can you tell us a little bit about what this journey has taught you about what actually makes these tools stick, what gets people to trust them and really use them in the flow of work?

[00:18:22] Speaker 4: Yeah, yeah, that's a really good question. Well, first off, thank you for your recognition of my promotion. I think it's awesome to work for a company that is seeing this landscape change and knowing that we need personnel in place to help us ride that wave and ride it well. To get to a place where we're not just observing AI and we're not just depending on the expectations that the vendors give us, but we're sourcing out real evidence. And so we're taking our stories from inside, internally, how our people are actually using it. And so I'm grateful to be in the role and help lead the charge when it comes to our AI initiatives.

[00:19:04] Speaker 4: In terms of what makes things stick, wouldn't it be nice if it was just an impressive tool?

[00:19:10] Speaker 2: It never is, though.

[00:19:12] Speaker 4: It never is. And Rich, I've heard that story so many times. The CEO of StackOverflow recently told us that adoption is not a technology problem, it's a people problem. So tell us, how are you making it stick?

[00:19:26] Speaker 4: Yeah, so I think the best way to look at it is to actually take a couple of cues from what we just heard from Derek. Derek said, I didn't even touch the tool because I didn't know if it was safe. So we wanna make sure the tool's useful, right? We wanna make sure that it addresses real problems. People have real efficiency gains. Things are taking less time. They're feeling like there's a wind at their back in a sense when it comes to doing work.

[00:19:52] Speaker 1: In a sense when it comes to doing work. But there's also this trust piece. And I think we have to really recognize a lot of the baggage that people are coming in with when it comes to AI. No one wants to be liable for some kind of data leak or anything like that. And so it's on the company to really say, hey, this is the approved tool. This is what good use looks like with the tool. This is what's appropriate to put in the tool. And if you hit a roadblock, where do you go to get help? All of those things to me create, they're the ingredients to the trust cake, so to speak. What I realize is that adoption isn't happening at the point of access. It's happening at the point of trust. And so when people trust the tool, trust the company they work for, and if we're really honest, trust themselves, right, because they need to feel like this is something that's empowering for them, not something that's bearing down on them, right. And so once we have those ingredients in place, that's really the red carpet for adoption to strut right into the organization.

[00:21:02] Speaker 2: Yeah, yeah, totally. I mean, who has time to invest in learning a new tool or a new system that you don't feel confident you're going to be able to increase your fluency really quickly and it's going to have an impact, so that makes a lot of sense. Wow. So NASCAR, a total legacy organization, has lasted more than 80 years because the sport is constantly adopting new technologies. So where are you actually seeing AI show up in day-to-day operations today, Rich?

[00:21:31] Speaker 1: Yeah. And I think that that legacy piece is really important because it means that this isn't the first time that NASCAR, as an organization, is facing a changing technological landscape. We remember the Internet. We remember e-mail, social media. All these things were at one point a novelty, ones we weren't sure about. We didn't know if they were going to stick around, but ending up changing the whole way we do work. And so this is our current wave. And where I'm seeing it show up day-to-day for people is right in that foundational zone. They're finding the information faster. They're synthesizing that information faster. We're getting first drafts at a speed we haven't seen before. People are able to get faster summaries. And at first, when you hear that, you're thinking, okay, well, these are more fundamental or rudimentary maybe workflows. But what I've come to understand is that this is the sweet spot, especially for people who are first engaging with AI, because this is where you have the AI aha moment. And the AI aha moment, I'm sure you and I have both had, where we sit down with the tool, we put in a prompt, and it blows our mind. It showcases something we didn't even know was possible. And the faster we can get people to a place where they are in awe of what the tool can actually do, now we're starting to get the hamster in the wheel running, and the gears start turning, and people start thinking about innovation.

[00:23:10] Most of the use is right there at the beginning. I have a blank slate, my brain is fried, I'm scattered, I have a meeting in five minutes, I don't know what I'm going to say. These are the use cases that we don't talk about a lot, but guess what? These are some of the most relatable across, not just sports, but every organization. And so, to me, I'm not looking for the shiniest use case, I'm actually looking for the most repeatable use case, because, to me, the beautiful moment with AI happens not when it's like this thing we observe, but when it really starts to become a part of your life and starts moving the needle on your work.

[00:23:48] Speaker 2: Totally. Like a tool that you can't imagine living without anymore. I started at OpenAI three years ago, and Chat GPT was just released, so I really wasn't using it yet. It was not commonplace. I remember I couldn't even get access to it until I talked to a colleague, a friend that worked at OpenAI, and I was like, can you please get me access to this tool? You know, it's all the rave. And now, three years later, I just can't. I can't imagine doing my job or even living my life, the personal day-to-day things without it. And I find myself... I don't know, have you heard of your colleagues, your staff taking it home and learning at home? I find myself in bed at night, like practicing with codecs and projects and how to make custom GPTs, because then you really like the curiosity. It starts to lead you. And you're like, wow, I know that this is capable of even more than I'm doing right now.

[00:24:50] Speaker 1: More than I'm doing right now. Have you heard of any of your colleagues getting excited and really getting creative with the use cases?

[00:24:57] Speaker 2: Yeah, it's funny. I was talking to a colleague in information security and obviously we're talking about really nerdy IT things, but then he shares with me how he's using AI to plan vacations at home. I love hearing that because to me, the more at bat, so to speak, you get with AI, it's only gonna make you better. And so whether you're using that tool at work or at home, you're still building competency. The more you integrate it into your life, I think the more you're starting to extract the value.

[00:25:34] Speaker 1: Absolutely, awesome. Well, you guys are definitely on the right track. As I told Derek, you're actually light years ahead of lots of legacy enterprise organizations. You guys are just blowing my mind. So when you talk about building an AI operating system, not just deploying tools, can you tell us a little bit about what that looks like and what have you learned about what actually makes that system work?

[00:26:01] Speaker 2: Yeah, I think an operating system is necessary because we need a full path, a full handhold, I believe, from access to impact. We can't just hand people a tool. People need to know where to go. How do I access the tool? Is there SSO integration here? That's one. Two, what is approved? And not just approved from a governance perspective, but also what's approved in terms of this particular workflow because it's easy to get lost when you're using such a powerful tool if you're not able to aim it correctly. We also want efficiency in terms of how we use it. Are we using it well? Where the boundaries are? What should or shouldn't you do when accessing the tool? And then where to get help, of course. It's important that we're creating clarity and confidence in our people when it comes to AI, that the next step is one that we've anticipated. And so it's not just an overview, like, hey, let's send Rich an email to see what he thinks. There's a system that every single AI initiative, idea, dream goes through to make sure that it's safe for the organization, and it creates speed. It's safe and speed. And so that's not something that is ad hoc; that's something that is built right in from the ground up. We've created this AI front door, so to speak. I kind of refer to the process as almost like having an AI bouncer. It’s like, whoa, okay, AI initiative. Let me check your pockets here. Are you trading on our data? You know, that kind of thing. As soon as that AI tool crosses the threshold, it morphs from being a bouncer to being like a tour guide. And so it's a place where now I feel safe introducing, like, oh, okay, you're safe. Let me introduce you to the HR team. Let me introduce you to the legal team. Because now that I've assessed safety, now we're going to move towards speed. Now we're going to move towards helping people. To me, that has to happen at scale and it has to be repeatable with a level of consistency. You can only get that when you have an operating system.

[00:28:25] Speaker 1: I think we need to definitely translate this talk into a one-pager. We need to make sure that we pull out some of the cool analogies you've used, like the AI cake, the AI bouncer, the AI tour guide.

[00:28:41] Speaker 2: Right, right. This is how I think. Oh, it's very fun, and it makes it fun. Obviously, your disposition as a leader is in part why people are trusting you along this process. It goes a long way.

[00:28:54] Speaker 1: Okay, so we're all very interested in this next question, and you have been so successful in this regard. You shifted the conversation from automating repetitive tasks to reducing friction across all work. How has that changed the way your team thinks about AI and how has it impacted concerns around job replacement?

[00:29:14] Speaker 2: Yeah, so let me say this super crystal clear. At NASCAR, we see AI not as a robot in your chair but as a jet pack on your back. I take the time to make that clear because I am not interested in a battle between AI and human beings. I'm seeing the value of what happens when the AI and the humans collaborate. That to me is the strongest force. It beats AI, and it beats the human, right?

[00:29:48] Speaker 1: The human, right? The combination of the two. We are greater than just the sum of the parts. The work that I've just been seeing throughout the organization, when human ingenuity meets an acceleration pedal like AI, it's incredible what we can produce. And so I think when we talk about friction reduction, that really changes the conversation. I think the word automation can trigger some of that replacement fear. But when we talk about reducing friction, we're saying we want the slow work to become fast work. We want repetitive work to not drain you, right? We want you to have more of your brain capacity freed up for human judgment, creativity, more executive decision-making, right? This isn't, is this replacing me? It's how can this help you work better. And so the better I can make that jet pack and the more backs I can strap it to, I think how we're all gonna soar.

[00:30:48] Speaker 2: Yes, Rich, I love that. And also your approach really illustrates how much words matter, how much language matters and the way that you describe new systems and new approaches and new tools to each other and to the people that you're leading. I really, really love that. It's a friction reducer, not necessarily, and I think you said not an automator. It's a friction reducer, I love that. Man, your team must really love you.

[00:31:19] Speaker 1: I hope so. I think that we've talked about this a little bit, but we're gonna host an in-person event for all of our friends working in sports and the athletes and sports adjacent, and I hope we get the opportunity to meet more of your teammates.

[00:31:36] Speaker 2: Yeah, I love that, absolutely. So for leaders in the audience who are earlier in their AI journey, what's one thing you'd recommend they prioritize first to build momentum the right way?

[00:31:47] Speaker 1: I think if you're early in the process, the first thing is, one quote I always remember is just, the best time to get in is when the tool came out and the second best time is today, right? So I think the first thing is to remove all sense of guilt because what guilt will do, it will cause you to press the gas when you shouldn't. And so you cannot come at this as a catch-up game. And so while I appreciate being able to be ahead of the pack in terms of sports and AI in some ways, I also want to charge others who are just getting started to not worry about who's in front, who's behind, right? This is you and your organization. The second thing is, it's so easy to get focused on the tooling, especially the more you're plugged into social media and YouTube and all that. It can feel like the whole game is about what tool you pick. And the real thing is, what are you gonna do with the tool? And so statistically we see like lottery winners end up bankrupt, right? And so what we're seeing is that an influx of capability can actually be the thing that is your downfall when you're not prepared for the capability. And so what I would do is dig deep, get a good foundation, leaders, really think about what you want from AI, have the picture of what a win is before you ever bring in that tool because if you don't, then you're really just shoehorning AI and you're not actually planning for it. And so what we've seen is that the more of a foundation you have, the better governance you have in place, having legal review, making sure the tools are safe, making sure that people know how to access the tool, where everybody doesn't have to recreate the wheel, you're extracting some use cases, playbooks from some of your top-level AI workers. And so I think that's where I would start. I would start by building a great foundation, having a good plan and knowing what a win looks like for AI before I ever considered a this versus that conversation.

[00:34:03] Speaker 2: And it sounds, Rich, like maybe you can lean in on some of your more technologically inclined or more innovative teammates to help you identify what the wins look like. So I know at every company, you have early adopters who see possibilities before everyone else. So how have you empowered those people while still making sure the experiments scale responsibly across the organization?

[00:34:30] Speaker 1: Yeah, and so the goal is to make sure that those individuals are recognized. And I think that's the important, it's super important to make sure that people realize that the exceptional work that you're doing is one that we see and one.

[00:34:46] Speaker 1: One that we see and one that we value. And we're not just turning our best and our brightest into places where we just extract value, but it's also a place where we pour in recognition, right? But another thing that we can get from those people, they help us see around the corners. They have the use cases that maybe the rest of the organization isn't thinking about. They carry a lot of the innovation that if we aren't extracting, then we end up having like this very popcorn style innovation happening at the company where it's person-led, where the people who are the most tech-savvy, who have the most AI experience are doing 2x, 3x, 5x what the other employees are doing. We're not extracting the value. We're not showing people what those folks are doing and how they could start implementing some of those as well. When we're not sharing the knowledge, it really kills innovation at the company. And so it's important that we highlight those early adopters and also help to extract the value from what they're seeing and then deploy that to the rest of the organization so that we're able to keep up. We're not recreating the wheel every time with a new employee.

[00:36:05] Speaker 2: Well, Rich, it honestly, I keep hearing, I heard you and Derek talk about recognizing people, making sure that people are seen from the fan side of your work to internally. And I think that's so meaningful. The audience is obviously signaling that they do too. I just wanna read a comment from the audience here. David Christie, manager of enterprise data stewards, says he loves these takes from Rich. You need to take this show on the road. So many great insights and ways to look at this. So I just hope that you guys take that with you and know that you're really moving us today. Thank you for sharing so much. So one last question before we dive into the audience questions. And it's just, when you imagine the workplace five or 10 years from now, what excites you most about the ways AI could expand what people are capable of doing in your organization?

[00:37:03] Speaker 1: Yeah, I mean, that's an eternity in AI years. It is. But outside of my Jetson-like dreams of what things could be, I think the thing I'm most excited about is that you can actually have modeling, you can actually have artifacts that you've created that are being done at the speed of your own ambition. And so the more we can reduce speed around, or increase speed around moving from idea to prototype, to me, it allows people who maybe are further down the chain who don't have the time or the resources or even the experience to be able to see a dream come true, to be able to see an application that they thought of in their bedroom or coming out of a meeting and thinking, hey, what if we had this? And having just a couple minutes or even a few hours being able to showcase that. I wanna see in that 10 years, a lot more dreams coming true, a lot more prototypes, a lot more ways in which people who especially have been disenfranchised are able to utilize AI to close the gap in terms of innovation. So that's my big AI dream for 10 years from now.

[00:38:22] Speaker 2: Me too, Rich, I'm so with you. And you sound like what some of the most creative people I've interviewed in years, like you guys are in alignment. The most creative people, they're leveraging AI to prototype, to practice on their ideas before pouring resources into it, to leveling the playing fields in gaps in knowledge and experience. It can teach you so much, you can learn so much by becoming not even fluent in AI but just confident in sitting down and playing around. So thank you so much for all of that, Rich, and let's bring Derek back to the conversation now and take some of these awesome OpenAI Forum community member questions.

[00:39:11] Speaker 3: Hey, Derek.

[00:39:13] Speaker 4: Hey Natalie.

[00:39:15] Speaker 3: You guys are a really special team, so wow. I mean, I'm just very moved, you're obviously a very human-focused team. It's just a neat culture here at NASCAR. I would say just the support that Rich is talking about, having that kind of support from leadership to give us the confidence to try these new things is just cool, so I think that's very cool. So cool. Okay, question one from the audience. What signals tell you an insight is worth acting on versus just interesting?

[00:39:44] Speaker 1: Just interesting? And maybe, Derek, you could take this one because we talked about analytics a lot in our first segment. So I would say, that's a great, first of all, that's a really good question. Being able to take it to the group that I talk about every Tuesday morning, the insights knowing that's coming from various, at least two different data sets, help us, and especially seeing a pattern over time. Things may pop up from one race to another race, from one time to another time, but when we start seeing a pattern, that insight really builds. And so we look back on it, and we don't wanna make any, like I mentioned, knee-jerk reactions quickly, because anytime we make changes, like any technology, it's likely to cost somebody some money. And so we wanna be very careful about doing that and being methodical, but if it's the right thing for the sport, we will move in the right direction, make sure all the stakeholders are in the right position, in the right alignment. But at the end of the day, I think it really does come down to making sure that there's that pattern that's lasted over time.

[00:40:45] Speaker 2: Rich, anything you wanna add to that one?

[00:40:48] Speaker 3: No, that was great, thanks, Deirdre. Okay, awesome. So when applying AI to NASCAR, what was the signal that was most surprising? The insight or the aha moment that you experienced when you started leveraging AI?

[00:41:03] Speaker 2: I'll take that, just being a relative new, how easy it was. I mean, to get up and running, I almost felt guilty with how quickly, how much time saving it was. Like I mentioned, I went through a couple of the different training courses, which I recommend to anybody, just to make sure that you know how to upload data properly, just to make sure that you're given the tool instructions on how you want the outputs. For in our case, I would like to know the information in a technical manner. I wanna use technical language. And if there's something that's not favorable, give me some constructive way of explaining that to folks. But then also, just being able to have the right prompts. I think those are some important ways to go about it.

[00:41:42] Speaker 1: Yeah, I would also add, I think it's super important that we, I kind of think about AI, again, going back to the AI cake, like a three-layer wedding cake. And at the bottom level, we have that foundational LLM that we can deploy to most people in the organization. Then at the second level, we have, if you consider the first level, like a four-door sedan, then that second level is, well, some people might need a pickup truck and some people might need maybe a sports car, right? But that's really from investigating into that department. And I found a lot of the aha moments now are happening at that department level where it's like, oh, this is finance specific, or oh, this is like legal specific. And people are seeing like, oh, we can actually get deeper contextually. And that is creating a lot of buzz in the org.

[00:42:37] Speaker 3: Yeah, awesome. And what's one thing you challenge this audience to try with AI this week, something small but high impact if they're just getting started?

[00:42:49] Speaker 2: Yeah, I think I would do two, one at home and then one at your workplace. And I think one of the core things I find that people don't think about when they're using AI is bringing the problem to the AI. And so I find myself having to remind people like, oh man, I don't really know how to do this. Tell it, tell it I don't know how to do this. And be as vulnerable as you can so that the tool can really pick up the gaps and fill those gaps for you. And so I would really encourage people to do some consulting with your AI tool and say, hey, this is where I am. This is what's going on. Help me get my arms around this, help me get my mind around this, create a tool for me that might be able to help me track this better. Bringing your vulnerabilities to the AI tool would be something I think would be a cool experiment.

[00:43:50] Speaker 1: And then at home, I think it would be really cool. One of the thing I'm really passionate about because I come from the education side is that I think children should be able to get more touches with AI. I think we're preparing our kids for a world that doesn't exist yet. Part of this is gonna be competency. And so I think it's really an opportunity, obviously not to skirt the education system, but rather, hey, I wanna make sure that this tool doesn't get past you. I want you to be a digital native when it comes to this tool so that when you eventually get to a place where you're working, that this is kind of second nature for you. Especially for those with families, young kids, I would love to try that and just do a fun game. Make it tell you a story or something cool like that.

[00:44:38] Speaker 3: Yes, totally. You're reminding me of so many things, Rich.

[00:44:42] Speaker 1: First of all, I think that, you know, I have a 15 year old son about to be 16 and in the very beginning of Chat GPT, you know, he was ... the very first use case was he used it to write a paper, which he shouldn't have done. You know, he was in seventh grade. But then I increased my AI literacy and now we use it. I mean, it is such a powerful learning tool for him. You know, if we go to a museum, I'll create a scavenger hunt with Chat GPT first to help him stay engaged. I tell Chat GPT what he's interested in—he's interested in basketball and fashion—so help me create a scavenger hunt. That makes so much sense.

And then the second thing I just wanted to surface was yes, about asking Chat GPT because my family and friends, they all text me all day long, like, how do you use Codex? How do you do this? This isn't working. And I'm like, ask Chat GPT. Right. Ask the tool. So you are speaking our language and you are an excellent AI mentor, Rich. Derek, did you want to add anything to it?

[00:45:59] Speaker 2: I would just encourage anyone who may have, like, maybe it's a simple spreadsheet that has data in it, just to upload it and ask a question. For example, is there a relationship between column A and column B? Something simple. I think what you see as an output will really help you get on the path to really understand what this can do. Rather than push it off as, oh, this sounds scary, this sounds scary. I think that might be a good way to get started.

I love all of those suggestions. So, fellas, I'd like for you both to answer this one. What's one myth about AI and sports that you'd love to retire for good today?

[00:46:41] Speaker 1: I can start. I think a myth, whether it's sports or otherwise, is that AI is about pushing a button and it out pops the answer. Nothing could be further from the truth. We know that AI needs guidance, it needs coaching. You need to be able to keep an eye on it, right? It's kind of trust but verify. There are some opportunities to help guide it. I was mentioning the nut going through the bolt and getting it started. You got to have the right prompts. You've got to ask the right questions. And so, that's where the human part comes into it. After that, once you get the right prompt, then you can replicate it. In my case, I replicate after every single race so that my audience or leadership has the same orientation in terms of how we’re evaluating races. So I would say the myth of just, it’s all automatic, that's got to go away. There's definitely input and the human component that's necessary.

[00:47:37] Speaker 2: A myth that I would love to dispel is that AI work, whether it’s sports or anywhere else, is limited to just the funny videos you're seeing on YouTube and TikTok. So often, even just around my neighborhood, when I tell people that I'm working in AI, they're just like, "Oh man, I hate that stuff. It's just a bunch of AI slop." But there’s this whole enterprise version of AI, this ability to increase how we do work, how to engage our fans in a way that's more dynamic. As someone who's a fan of sports, I see this as a huge win. The beauty about sports is that no one is going to pick a bunch of robots playing basketball over watching the intensity that happens in the NBA.

I grew up as a nerd, as a gamer. I remember sitting in front of racing games and just watching the CPU race until I put my quarter in. That wasn't the fun part. The fun part was engaging with the artificial intelligence in the game so that I can make progress. Sometimes, the sentiment can be really strong about one aspect of AI, but AI is so much bigger than the talking food that you'll see on certain videos. We’re doing so much more analysis and drafting and workflows, even within IT teams, and seeing how AI is getting into the developmental workflows. It's so much more than just the visuals that you may or may not like.

When we say that we're engaging with AI, it's not just to show our fans stuff that they don't want to see. It's about getting the sport to a place where it's the most exciting it’s ever been, where it's a no-brainer to come to a race because it's going to blow your mind. And then when you're there, engaging them even more deeply because you can tailor the engagement based on what you know about them.

Yeah, that makes a lot of sense.

[00:49:40] Speaker 1: That makes a lot of sense. And I think you're touching on something rich too, that's really important to communicate and is a hill that I will absolutely die on. You know, as a community builder, as a former choral singer, as somebody who's not an athlete but loves to go to basketball games, you just love to go where people are. We really like to be around each other. And we learned that during COVID, during, you know, quarantine and school. Schools being closed down and everybody working remotely, we were dying to be together. In a lot of ways, companies like NASCAR are leveraging AI to just like enable us to be together in new, cool ways. So I just totally agree.

[00:50:28] We have one more question from the audience. This is from the Associate Dean of the University of Washington, Kevin Mihata. Kevin, good to see you, thank you for being here. He says, thanks so much to all of us beyond AI use within NASCAR. Can you see ways that NASCAR or other sports can help educate or advance fans' own AI engagement in some of the ways Rich has discussed?

[00:51:06] Speaker 2: Yeah, I think my hope is that these conversations don't stay siloed. Even though we're talking about AI in sports, I'm just talking about AI, right? And so my hope is that there are things that you can glean from the conversation that's more than just how it's applicable to sports. I think when we say that AI is a people problem, that means that it is a work transition piece. And so for all of my AI fans, NASCAR fans out there, utilizing AI on your terms, make it your AI. The AI story will either be told with you or without you, and I think it's going to be much more dynamic by having your voice as a part of it. So don't lose the opportunity to have your voice as a part of how AI is shaped everywhere.

[00:52:01] And so Derek, I really love when he's talking about how we really focus on the fan feedback because we're trying to make the sport the best they can for them. The more that they can engage and the more that they can learn about the tool, the more they can accept when we do AI initiatives; they can back it because they know that it's coming from a place that's for them. I think that's a great place. And also thank you for the great question. I have a great privilege. I'm here at the NASCAR R&D center, and I have the great privilege of often hosting students here at our facility, but we also get the chance to go into local classrooms and talk about what we do.

[00:52:26] Certainly, the drivers and the crew chiefs may be the most visible on race day on TV, but it's a good opportunity to talk about some of the other things that go on in the sport behind the scenes, one being data and analytics. I'm always careful to ask the teacher or whoever's in charge in advance if it's okay for me to talk about AI, knowing the sensitivities about how it may be used in the wrong way. They always, every single time I'm asked, say, yes, please talk about it because they know I'm going to talk about it in the sense that we're all discussing it during this chat right now. It's really leveraging data, using it as a tool, being more efficient.

[00:53:02] So that students can see, hey, I may not be a race car driver or another athlete, but I really like data, I like numbers. This is a neat way to help them get plugged in and see how they can use this, either in their high school career, college, or beyond, beyond just what may be obvious on a specific event day.

[00:53:31] Speaker 1: I love that. And if you don't mind, Derek, I said in the beginning, I hope this is just the beginning of our conversations and our collaborations. But we are also very interested at OpenAI in putting the tools in everybody's hands, increasing AI literacy, and hopefully just like really being a driver of prosperity. So we would be happy to continue to deepen this collaboration and even help the communities that you're especially students and teachers. I mean, we just love that. We really want to give them the tools and just let them run with it and shine.

[00:54:05] So let's keep that conversation going because we might be able to be of service. But that's a wrap, guys. Wow. What a fun conversation. I did not know how people-first NASCAR was before I was introduced to the two of you, but I think that your organization is very lucky to have you both and we've all learned a lot today. So, such an honor to have you.

[00:54:30] Speaker 2: Yeah, thank you, Natalie. Thank you to everyone who's tuning in and will be tuning in the future. Appreciate the opportunity just to share how things are moving with us and how we're really trying to...

[00:54:38] Speaker 1: We're really trying to drive AI adoption here at NASCAR, and I just thank you for the platform and the ability to connect with your audience. Well stated. Yes, everyone, thank you so much. Totally our pleasure.

[00:54:52] Speaker 1: So before you guys leave, just a few closing remarks. We actually have an upcoming event with our chief economist, Ronnie, and it's going to be about a new research report about understanding the labor market through real-world usage data. Rich, I think you kind of touched on this—that using it at home or using it in the office means AI literacy. It means you're increasing your confidence, your fluency, and your curiosity is probably increasing.

[00:55:27] Speaker 1: What we found is that the real-world use cases are quite personal. You know, the scavenger hunts at the museum, the calendar planning for your busy schedule of young student athletes in the house, the meal planning. So I hope you guys all tune into that because that will be a really fun one, and I think will give us awesome, more cool ideas on how to get started with AI in ways that are not scary at all.

[00:55:53] Speaker 1: We'll also be talking on April 22nd with some founders of Perturb AI and a member of our AI for Science staff here at OpenAI for Decoding Biological Intelligence, building AI agents for the brain genome. We are really, really focused on advancing scientific discovery so that we can make connections that benefit humans, and that's what we're going to be focusing on there.

[00:56:17] Speaker 1: All of those folks will be here in the office with me on April 22nd. And then if you guys missed it, on Monday, we actually hosted Sam Altman, Adrian Echofet, and our chief futurist, Josh Akyem, on building the future of AI. My colleague, Caitlin, has already published that in the OpenAI Forum. It's ready for on-demand replay. It was a very lovely, intimate conversation that I do believe will show you the more human side of us all at OpenAI.

[00:56:47] Speaker 1: The folks that are building this technology actually really care a lot about all of us and how it's going to be used, so tune in for that one. And you know, thank you so much for being here, Rich, Derek, it was an absolute pleasure. We will see you in June in Napa, and I hope this is just the beginning of our collaborations.

[00:57:06] Speaker 1: And you at home who tuned in, thank you so much for being just awesome community members. Thank you for all of the engagement, and we will see you again very soon. Good night, everybody.