DataTopics Unplugged: All Things Data, AI & Tech
Welcome to the cozy corner of the tech world where ones and zeros mingle with casual chit-chat. Datatopics Unplugged is your go-to spot for relaxed discussions around tech, news, data, and society.
Dive into conversations that should flow as smoothly as your morning coffee (but don't), where industry insights meet laid-back banter. Whether you're a data aficionado or just someone curious about the digital age, pull up a chair, relax, and let's get into the heart of data, unplugged style!
DataTopics Unplugged: All Things Data, AI & Tech
#60 AI and the Paris 2024 Olympics: From Tech to Yusuf Dikec Memes
Welcome to the cozy corner of the tech world where ones and zeros mingle with casual chit-chat. Datatopics Unplugged is your go-to spot for relaxed discussions around tech, news, data, and society.
Dive into conversations that flow as smoothly as your morning coffee (but don’t), where industry insights meet laid-back banter. Whether you're a data aficionado or just someone curious about the digital age, pull up a chair, relax, and let’s get into the heart of data, unplugged style!
Olympic AI agenda: Discover the grand vision of the Olympic AI agenda and how it’s set to revolutionize the games.
AI and tech innovations at Paris 2024 promise to be game-changers in sports.
Obscure sports fascination: Ever wondered why you can’t stop watching niche sports? Blame it on AI, as we dive into how it’s making you hooked on obscure sports at the Paris Olympics.
Yusuf Dikec memes: Delving into the meme culture surrounding Yusuf Dikec—because who doesn't love a good sports meme?
AI in gymnastics judging: Will AI be the future judge of gymnastics? We delve into the controversial world of AI gymnastics judging and what it means for the sport.
Zoe Hobbs' AI boost: Discover how Olympic sprinter Zoe Hobbs is getting an AI-powered boost to sprint ahead of the competition.
Track and field insights: Using a bird's eye view and detailed statistics to create a great visualization of the 100m final.
Recoloring history: The fascinating process of recoloring the 1924 Paris Olympics—bringing the past to vibrant life.
You have taste in a way that's meaningful to software people. Hello, I'm Bill Gates. I would recommend TypeScript. Yeah, it writes a lot of code for me and usually it's slightly wrong. I'm reminded, incidentally, of Rust here, rust, rust.
Speaker 2:This almost makes me happy that I didn't become a supermodel.
Speaker 1:Cooper and Netties. Well, I'm sorry guys, I don't know what's going on.
Speaker 2:Thank you for the opportunity to speak to you today about large neural networks. It's really an honor to be here Rust Rust Data Topics.
Speaker 1:Welcome to the Data Topics. Welcome to the Data Topics podcast. Welcome to the Data Topics.
Speaker 2:Podcast. Hello and welcome to Data Topics Unplugged, deep Dives, working name your casual corner of the web where we discuss all about the Olympics, ai and data. My name is Murillo. I'll be hosting you today and, joined by the one and only he is back. Hi, bart, yes, and behind the scenes, there's always Alex there She's's waving again. Just trust me, follow, just trust you on this one. Um, how are you doing, bart?
Speaker 1:I haven't seen you in a while very good, I'm uh happy to be back yes and I'm very curious to hear about what is laying next to you next to me.
Speaker 2:Oh, this whole thing. Yeah, so it's a hat okay the hat says uh.
Speaker 1:Paris 2024 yeah.
Speaker 2:So what happens in paris 2024? The, the summer olympics, which, coincidentally or maybe not, that's the, the, the topic of today. So, um, I was actually there, uh, this past weekend. So, yeah, today's was today's, the 6th of august, tuesday, so I was there earlier, like second, no, third and fourth, or, yeah, third and fourth of august. I was there watching the tennis finals. So, actually, like the way it works, you get a, you buy the ticket for the day and then on the day well, actually the day in the stadium, I guess, or at least that's how it was in the olympics and then you can watch all the matches there. So, so, watch the women's final. The men's doubles finals. Women's doubles finals women's doubles third place. Men's singles third place. There was a lot of matches.
Speaker 1:So with a ticket you get access to one venue, basically Kind of yeah.
Speaker 2:So the venue for that day.
Speaker 1:Was it easy to get the tickets.
Speaker 2:So my partner works for Bridgestone, which was sponsored in the olympics. So it's usually like a lottery system, but because she went through her work, it was like a smaller pool, I guess. So we applied and we got it and it's not that expensive.
Speaker 2:I think it was like it's not super cheap but you got the vip treatment no no, we actually set on one of the the last rows, but actually the stadium is really nicely set up. That, uh, there's no really bad seats. Okay, everything you can see everything really well, it's pretty inclined, so it was pretty good, and that must be true for the. So, for the tennis, uh, tennis fans there, the finals between djokovic and alcaraz oh nice yeah, and djokovic won and the atmosphere was really nice really nice so a lot of like spanish people.
Speaker 2:They're very passionate, so they're like screaming a lot, but then at the next point, like the djokovic fans, they were really like focus. It was really, it was really, really really cool. Um, and djokovic actually won. I think he's like the fourth player in history to win all the grand slams and the olympic gold medal. Oh well, so it was a big. It was a big uh, it was a big, big, big deal for him. So it was really cool. And also, um, a few things that caught my attention as well. One thing that they did actually like in between, they have the big screens right and they had, so the 1924 olympics was also in paris oh nice, 1924, 1924, so 100 years apart.
Speaker 2:Yeah, and they were comparing like how the olympics were in 1924 and olympics now, um, and sometimes they'll have like these videos that was like just pictures in black and white, but then they will be playing songs and then they had, like you know, I think there's like apps. You know that you take a picture and then it pretends like the photo is singing.
Speaker 1:Okay.
Speaker 2:So they will do that with, like some popular songs to just kind of get. So I thought it was pretty cool Like you're using ai, you know.
Speaker 1:nice also made me wondering you know, like uh, what else are they using ai for? It's like for the topic, there is a topic of the day.
Speaker 2:We want to have a bit of an open discussion on uh ai and the olympics yes, indeed, indeed, I think, uh, and actually, well, maybe is there something that really struck you a lot, like anything very surprising when it comes to ai in the olympics or in the olympics in general well, maybe both, but maybe we can start with olympics in general.
Speaker 1:Then we can go to to our topics um no, tell me, I think you have something in mind.
Speaker 2:Tell me no, no, no, not necessarily I'm not a very uh.
Speaker 1:I'm not a super dedicated olympics uh viewer yeah, but you like some, what are you? Like I follow some, uh, some like I like to the follow the cycling for the triathlon a bit uh, swimming a little bit, don't you want cycling no uh, yeah, I'm going to pullel won both the time trial as well as the road race. There's going to be climbing tomorrow. The whole athletics stuff just started up. I'm following that a bit as well, but I'm not like I'm going to make time in a day to follow it.
Speaker 1:If there's something interesting or an athlete I'm following.
Speaker 2:I might do like an evening rewatch or something yeah, I think there's also so many sports right it's hard to really keep up and like what's gonna be when and there's this like it's. It's a lot of stuff. It's a lot of stuff.
Speaker 1:But yeah, and in regards to to data, specifically, like uses of data and ai well, I must say that in the news during the olympics I don't think it's really like a point, a topic that yeah that um is really front and center. Yeah, but I did google, did google a little bit and I did not know, but apparently there is a.
Speaker 2:There was an olympic ai agenda yes, I also ran into that, but what is it about?
Speaker 1:That was, I think, presented in I want to say April. I'm not 100% sure.
Speaker 2:I also have.
Speaker 1:And basically you're putting it on the screen. Yes, so for the people watching, it's basically they emphasize that AI in general has become a big thing and they want to set out some guidelines around this, and they have five focus areas. Maybe go over them very briefly. One is the supporting athletes clean competition, safe sport. One is the supporting athletes clean competition, safe sport and that's a bit like how do you use AI for talent identification, for personalized training, injury prevention, anti-doping efforts, these type of things. Second is ensuring equal access to AI benefits. I think that's a good one. You see this also you can make a parallel with other guidelines as well. For example, in cycling, you have this rule that you can only use material or bikes or clothing that are available to the general public, so you can't use something that would give you an unfair advantage because someone else cannot purchase it, for example, I see, but then it doesn't discriminate on the price necessarily.
Speaker 2:It's just like stuff that you an unfair advantage because someone else cannot purchase it, for example, I see, but then it doesn't.
Speaker 1:It doesn't discriminate on the price necessarily, like when they say just like stuff, that price can still be I can still be yeah, so like something super, but it has to be available to the, to the consumer at large. Okay, okay, interesting, um, so that is one of the points ensuring equal access to ai benefits. Uh, there's also the operations point optimizing Olympic and Paralympic Games operations, so reducing costs, making decisions around event operations more easily, improving sustainability efforts. The fourth one is growing engagement with people, both with athletes as well as attendees. And the fifth one is, uh, driving efficiency in general. Uh, I think, financial management, talent acquisition, onboarding of the ioc the olympic committee yeah, I think, uh, and this is all.
Speaker 2:Yeah, I think it was. It's interesting. I also came across that and I was also looking at how they're actually applying this in the olympics, like how this actually comes to be, um, maybe I can actually pull it up, let's see if I can find. Is this the one? Yeah, so yeah, in tech innovations in paris 2024. So they also talk about the, the, yeah, the ai agenda, and yeah, they talk about optimizing for sustainability, knowing kind of projecting, what's the when places are going to be super crowded, right, uh, also about, um, marketing, right, so giving personalized feed for you based on what you want to do.
Speaker 2:Uh, I thought it was very interesting, but it also like we, we work in data and ai, right, and I think a lot of these things are things that we've seen at, uh, different companies, right, which come to me is like yeah, you're applying ai, as like the olympics is an organization, as a company they're applying in different fronts, yeah, and when you look at that, it's like it's not for me. I mean, maybe, maybe I'm biased, right, but to me it wasn't like wow, this is so innovative, right, like I've seen these things in other places, like predicting this or sports analytics and all these things, right well, maybe what it links a little bit too, because that's a little bit of the the thought I had with reading their olympic agendas, like they don't really define ai I have the feeling like what they're talking about is like anything to do with making decisions.
Speaker 1:Making analysis, making conclusions based on complex data is under the header AI.
Speaker 2:Yeah, yeah, yeah.
Speaker 1:I have that feeling a little bit. Like there is just purely exploratory analysis, there is traditional machine learning, there is the stuff around LLMs actually being implemented, but everything gets grouped under all of this is AI.
Speaker 2:Yeah, actually, I think I saw something Like you mentioned LLMs. All of this is AI. Yeah, actually, I think I saw something Like you mentioned LLMs that they wanted to build a chatbot for athletes yeah, exactly, yeah, like with FAQs and how they can quickly get answers to their questions about doping, about various, about all these things, which I thought it was pretty cool. Let me see if I also. One other thing that caught my attention about this. Let me see if I can. One thing, one other thing that caught my attention about this, let me see if I can find. Is this the link, the Olympic movement, olympic AI agenda, and I think there was like a committee, you know, let me see if I can find this, because they brought yeah, I think this is this Anime session, ai sports. I think this is this 9-Minute Session AI Sports and this is the attendance. So I think they basically got like experts from different fields right.
Speaker 2:Ah, the working group. Yeah, so what is it called? Olei, olympia Agenda? Okay, all in all, there is a working group that defines these policies, right?
Speaker 1:AI.
Speaker 2:Working Group. One thing I thought was interesting is that there's a familiar name here. Oh cool, I didn't know. Yeah, so maybe for the people listening, we have people like professor of electrical engineering, computer engineering, at Purdue University Actually, that's the school I went to in the US. Well, it's that big system, right? So this is the main campus. But one of the people here, jesse Davis, professor Department of Computer Science at KU Levin, so he's the one that is very into sports analytics. So even when we talked about sports analytics, the algorithms, all these things is all under the guidance of Jesse Davis. Oh, that's cool, so very cool. And I think I looked at some pictures afterwards and I could see him in front of the stage there with the Olympic stuff.
Speaker 2:So I thought it was pretty cool, kind of give a little shout out to k11 and jesse, maybe one day we'll get him here on the podcast and we can, you know, corner him like.
Speaker 1:So I think this is jesse davis here, so cool, um, I tried to dig in also a little bit. Like what is it? Like? What did this translate to this olympic agenda? Like what does it actually mean, uh, for the olympics today? Um, one of the things that is interesting is that they use AI apparently it's Intel doing it to identify interesting highlights, and apparently that's the article by that we'll put in the show notes, the article from yahoo maybe you can open it uh and it's uh.
Speaker 1:It shows that they are using uh, intel's ai for whatever. Whatever that means, like their ai services to go to a huge amount of footage, all of the footage of the Olympics, to understand a bit like what were the main sensational points, and apparently one of the key variables is the crowd noise, like do people start cheering or not, and that they really push these in YouTube shorts, tiktok, twitter, like a short format, and that's why we're seeing, let's say, us as an audience in general, like much more less known sports than we used to.
Speaker 1:Ah, I see, Popularizing like yeah and just because we only take away what is really interesting and that's probably then easier to hook yourself on than yeah trying to immerse yourself fully in a complete day of sports that you're not familiar with or no interest in that's cool.
Speaker 2:That's cool and I do think olympics they kind of embrace a bit that novelty.
Speaker 1:You know like there are sports that are following the olympics that you never follow, and this is actually like for those rather unknown sport, very, very valuable right to surface like this, yeah I completely agree.
Speaker 2:I think he kind of talks you mentioned one of the the principles there was like about democratizing a bit like sports and talent and all these things right, bringing lights, putting the spotlight on the people that normally wouldn't. So I think that kind of falls there very cool and I think maybe on that as well, um, some sports that people don't follow or some people that are maybe wouldn't catch as much attention, maybe think of some memes that I saw.
Speaker 1:Um, maybe I'll give some context before you know this guy, uh, it's always good, when you start about memes, that you have a bit of context so you know this guy, this is a, so it's a shooting, it's like um are you showing like a he's a?
Speaker 2:turkish guy. He's a yusuf uh deke, I don't know, probably not saying uh his name correctly, but it's like about shoot, like we have basically. I mean, I don't know what's the actual name of the the. How do you say it actually shooting target shooting?
Speaker 1:I don't know Something like that.
Speaker 2:But basically you have a special handgun and then you just have to shoot at the target, right, yeah. And then on the left you see someone from maybe Denmark, maybe, that has headphones, you know like noise-canceling eyeglasses with super high-tech-y stuff, you know like different lenses and the gun there. And then on the right you have the turkish guy which looks like he just woke up, you know like he just drove the olympics. You know he doesn't have anything. He has a turkey uh shirt, right the uniform, but he has like regular eyeglasses and he's just, he's just shooting, um. And then you see, like on the video actually, that he actually does way better than this guy. So, and they made a lot of memes like let's see. So you see here like almost dead center, and the guy is a bit off. And then he really stood out as this guy that he really doesn't care, he just goes there and does his job, um, but he's doing much better. He actually got he was a silver medalist and everyone else had like super expensive headgear glasses and all these cool.
Speaker 1:There is this uh, this uh, uh. Well, it's part of a bit of a part of the meme that the uh turkey rang up something from someone from the secret service and said, yeah we need someone at the olympics here. Can you join us?
Speaker 2:yeah, exactly like when you were a retired hitman right, like he's like yeah, so there's a lot of stuff that uh came up, you know. So this is a famous tweet as well like, like data scientist, and then you have the guy with all the headgear and then a counter with Excel, and that's the guy, and you know that actually nails the bullet.
Speaker 2:Exactly, that was the bullet and the same thing here, like a rookie data scientist with like import, a circular and TensorFlow towards Keras. And then you have another person, senior right and it just works.
Speaker 2:Yeah, and it just works and the last one is the same two people. But then the guy uh, turkish guy is uh has a title like I got a new job. And then on the bottom one is like I am beyond thrilled to announce that I have landed my dream role ever since I was a baby. My destiny and hashtag leaving the dream, you know, like the chagi pt answers, you know so I thought it was uh, yeah the internet was very quick to reuse this as a meme and things like that.
Speaker 2:that, I think, gets a lot of spotlight, have a lot of attention, but you would never see this anywhere else outside the Olympics.
Speaker 2:So I thought it was pretty fun. I thought it was fun. Another thing, too, that I thought this is actually maybe the most interesting thing that I've come across, and it's not exclusive to the Olympics, but it's the use of ai and data in sports in general, I guess. Well, not in sports. In gymnastics, they are using this in the. They did use this in the olympics. I want to say it's not very transparent, so I'll already say this right now. Um, to summarize a bit the question I guess I found here on Reddit as well, they were trying to discuss a bit Will AI be the impartial judge of the future, right? So they're talking about how, in gymnastics, very small errors from the athletes can have a penalty that actually costs them the tournament, right? Like, very small, like you have to, I don't know you have to, I don't know you have to. The split, then the leg needs to be at 180 degrees, but if it's 175, you already penalize a few points, right, and all these things happen very fast.
Speaker 2:And also there is a bit of the, the bias, the human bias right I also heard another podcast that gymnastics is also a bit of a special one because it kind of holds a bit the geopolitical influence of countries, right, like, judges have a bias as well, and if there is a new athlete that they've never seen before and even if that person does well, there are there's an argument that judges are biased against them already because they don't expect much from to begin with, right. So then there is a bit of the question can we use AI to be the unbiased judge on these things? And then on that, there's a really very interesting article from MIT Technology Review, this one here. So for people that are interested in AI in gymnastics or just gymnastics judging, I would recommend to take a look as well. This is actually in the 2023 World Championships, which is actually in Antwerp, so last October, so last year here in Belgium. Basically, they're saying here that this guy, he did really well, but he got a disappointing score, and then you can actually appeal. The difference is that the appeal usually is someone that they review the tape they replace, right, but this time it was actually an AI system Maybe I'll go very quickly here, because there's a lot to unpack, right.
Speaker 2:But then basically they say that? Well, first, gymnastic is subjective, right? It's not something like there's how you dress, your facial expressions relating to the audience, the outfits, right. So there is a bit of the artistry of gymnastics, but then there's also a bit the objective part. You know landing, for example. If you need to land and you take five steps because you couldn't balance yourself, that is a very objective way to measure that you didn't perform well, right, um. So they even talk about like, yeah, like what's perfect also changes over time. People become more strict, right, um, maybe also to yeah.
Speaker 2:So basically, what uh, actually people play this video and talk in the background. What uh they were showing is that fujitsu technology company. They would take cameras, high definition cameras, from a lot of different angles. That is already part of the stadium, the setup, and then they reconstruct 3D bodies of the athletes, okay, and then from that you can actually see did they actually do a hundred, three, 720 degrees rotation or whatever? Did they actually land right in the middle? Did they do this and do that? So, in the middle, uh, did they do this and that? So in the video here you can gonna be able to see some of the, yeah, some of the simulations, um. So that's not really ai right.
Speaker 1:That's just technology in sports, I guess well, that's what is in what's in the name, right? Exactly, but they did also mention that because we see this in all the sports as well, like where we have uh with with uh, football, soccer, yeah, yeah, where we also have, like today, a lot of detailed video footage where, uh, where there can be a review based on these, on these footages, to see if something actually did happen or not happen, like was there a file, was another file yeah, indeed um so I think, uh, yeah, indeed think this is.
Speaker 2:I think it's interesting to bring it to this context, right, because, well, and also to elaborate a bit more on how, why is it so hard, why you need this support, something that I wasn't really aware either. For example, they mentioned this. How they call it here, it's a performance switch ring leap, right? So basically, the legs Switch ring leap.
Speaker 1:It looks like something. If I would do it, it would be the last thing that I do. Yeah, exactly.
Speaker 2:That's it. If the bottom used to be an ambulance, you just jump in and just stretch the legs. So basically, there's an athlete, a female athlete that she has. She's basically doing the splits midair with her back arched, almost touching her feet right, and then they say, like they actually talk here, like people usually avoid this because it's notorious to be downgraded in the difficulty score. That's because judges are especially strict with it, according to the code of points.
Speaker 2:So that's their manual right, for to get a full credit, the upper back must be in an arch and the head released, the legs must reach 180 degree split, the front leg must be horizontal and the back leg bent, with the back foot reaching the crown of the head or higher. So, and all of this that you need to take is in a second under a second right. So these are the judges that look and just kind of write some stuff down. And this is the point that they're saying yeah, this is happening, so there's going to be human error. There's way too much to look and the details are too minute to really rely purely on human reactions. And there's the whole bias part right that they mentioned afterwards here, this athlete I think she called it letter bias right, theotone bias.
Speaker 1:yeah, Theotone bias, yeah.
Speaker 2:So basically, if you're not a known athlete, people are going to expect you to score lower and that's going to be a bias as well, and that's where I could help.
Speaker 2:But, interestingly enough, this athlete also mentions how she does really well, so I forgot the name, but it's basically it's not like the rings or something, it's more like a bit of a more artistic, like dancing, kind of routine that they have the ground, they do some maneuvers uh, alex, if you know the name of jumping, I just. But then she also mentions how she, she likes that or she performs better than that, because she also makes eye contact with the judges. She has facial expressions, there's a bit of a like almost acting, you know, to it, and that is part of gymnastics today. Right, and that is something that ai would not be able to replace at all. Right, you cannot look into the camera and just like, make facial expressions and then hope that you connect with them or build rapport and see how you're doing real-time sentiment analysis yeah, exactly, maybe, yeah, it's a smile, okay, it's like, yeah, how happy is that person you know?
Speaker 2:um, so then there was all like basically, this article discusses all the different implications and they mentioned that there is the ai judging system. It already exists, it's already in place, but it actually doesn't fully judge the, the athletes. It's only on cases where the superior disagrees with the judges okay or when someone appeals um, and even then it's not used autonomously, it's used as a tool to support the the judges to see what happened basically yeah, exactly, uh.
Speaker 2:Maybe another thing as well that I forgot to mention the actual ai system. It was trained to detect different skills, different movements. Um, how much time do you have to wait to say, okay, this is one movement versus, this is two movements, and also score the movements based on the 3d simulations? So they actually it is a full blown uh judging system. It could be autonomous. I think in the article they mentioned that uh 90, the same as what the accurate uh, the judges would rate the movements right, interesting and the different variations.
Speaker 2:But even today, this is not used as this, it's used as an advice, and they even mentioned that this is not transparent. So this guy this is the guy from the beginning of the article so he actually appealed, his score went up 0.2 uh and he actually got the silver in the world gymnastics, uh, but he didn't even know that they use the ai system. He actually had to ask and then they said, yes, we actually use this to support our, our scoring. So it's not very transparent to be seen what would be transparent. And actually, fujitsu, the company that created this. They already mentioned that what they want to do is to commercialize this, so they want to sell this for different athletes so they can kind of have a home as a training.
Speaker 1:Exactly, basically, exactly interesting exactly so I think that is a bit like like you have these bit of two aspects to this. One is like really data gathering to see what happened. The other part is making conclusions based on that yeah, exactly that is a bit like how far? Can you automate that? Because you can make conclusions if you're 100 sure that the data that you're looking at like is it correctly interpreted, because here interpreting moves um, so do you have potentially a risk there, like that the interpretation of the data is not 100 correct?
Speaker 1:like yeah did because you have, like, probably, hundreds of data points, thousands of data points for a movement, if you're, if you're monitoring this, this that you want to summarize to something that can be interpreted yeah and then, based on that, you want to make conclusions yeah and that's like there are a lot of steps, steps. You need to be sure that you're.
Speaker 2:And I also think that the artistry part you cannot really replace, you cannot automate, I think. I think there is a bit, In a way it's a bit strange because that's a place where I guess bias, in a way it is desired. You know what is a I don't know? Because it desired. You know what is a I don't know? Because a bit of matter of taste, I guess.
Speaker 1:Right, it's not something you can really quantify in numbers and be super objective. Yeah, yeah, I think um, most likely if you have a, if you assume that you have an ai system that judges this, yeah, and you test as well that there is less bias than some random judges. True, right, true, yeah. But then you're saying it becomes a two mechanical judgments. More or less that's what we're saying like yeah you lose a bit of the personal touch. The question is, how desirable is the personal touch?
Speaker 2:well, according to the athletes that were interviewed or that mentioned, the coaches are athletes. They did say that it that's kind of bit what gives the the flair of gymnastics. So if it was too mechanical they wouldn't yeah, it wouldn't be appreciated. Yeah, but yeah. So I think in that way it's like you want it to be more objective because it's fair, but on the other hand, you don't want it to make it too mechanical but that's here for specifically for gymnastics, because there's like an artistic component to it yeah well, if you would discuss something like football yeah indeed, probably everybody will be happy if you have a system that is truly correct.
Speaker 2:I think so. So for football, I think so, but at the same time I do think we could be further with technology than like, even like, for example, the VAR thing, the 3D, like there were big discussions right before it was approved and I personally don't resonate. It doesn't resonate with me Like why would you not want this? But I do think there are some people that don't want it, that they still want to have a person there, they still want it, like otherwise it's not really football.
Speaker 1:You need to have someone to complain about.
Speaker 2:Yeah, I guess Like you need to appeal. You know you need to appeal, you know you need to convince them.
Speaker 1:I don't know, it's like it's a bit weird to me, but yeah but you lose a bit like this, this human factor in, it's like yeah, yeah I guess it's like you do lose, but to me for football it's like the human factor this maybe for football.
Speaker 2:There's like too much at play yeah, but I think it's like the human factor is like it's actually the players right, like the referees is just a means to an end. That's where I get to, even tennis.
Speaker 1:Tennis is something that you could fully automate if you wanted. So you could maybe with football. Now the VAR is like the I'm not a football expert at all, it's used as a review. Something is not clear. It's not the default. Even then, it's not like an algorithm. Algorithm, it's just a whole bunch of cameras and there are people behind it.
Speaker 2:Yeah, exactly, yeah, but like, but there's people like, it's just that now people can look stuff through different cameras that's it, so you get a full picture of what happened basically.
Speaker 1:Yeah, but even when you try to do with it as well, right?
Speaker 2:yeah, but even then you don't use vr it. It's only used for certain key moments of the match Most of the times, it's not.
Speaker 1:Okay, okay, but in theory you could move it the other way, where the default becomes an algorithm that judges. Does this happen correctly? Yeah you could. But from the moment that this algorithm takes a decision like there is a file, there's no file like a decision, moment that you have an actual ref verifying the decision but it could.
Speaker 2:I think today you could have like a group of four or five reps behind screens just making calls. But I think tennis, you could go even further. I think tennis you could almost fully automate. You know, I think tennis, yeah, the ball goes in, the ball doesn't go in. Did it touch the line? Did it not touch the line? We already have systems for this.
Speaker 2:It's like it's very yeah, but we still have an umpire, we still have line referees, whatever. I think the only thing that is kind of subjective and that's something that I realized this past uh, tennis match that I want to watch is usually tennis players. They have 25 seconds to serve, yeah, but a lot of times people are hackling, you know, they start screaming and stuff, and then usually the umpire gives them more time and they pause, and I think that's the only part that is a bit subjective, but everything else really is very black and white. Did it touch the line? Did it not touch the line? Did you go this? Did you do that? You know, so it could be automated, but people don't. It's not on top of those minds, I guess.
Speaker 1:And maybe when we discuss collecting data to interpret. I think that also, if you look at the first point of the Olympic agenda, it was to support athletes.
Speaker 2:Yes.
Speaker 1:I think we have an article about Zoe Hobbs, olympic sprinter, zoe Hobbs gets AI boost. I think it's a good example of a way that ai that's that's how they call it it's used to to support an individual at least as well, where they um have a tool, in this case by uh view motion labs, um that analyzes images of uh zoe sprinting zoe is a is Zoe is a sprinter and to analyze, like to do a gait analysis, things like these and to either improve gait or for injury prevention, these type of things and in the end, you can discuss what part of this is AI or is this complexity analysis or whatever, but what you see today is that these type of tools bring this to the broader audience.
Speaker 1:It is much more easy for your average good coach to have this type of technology than it was 10 years ago. It probably existed 10 years ago, but today this is much more achievable. To purchase the technology, to have it in hand, but also to draw conclusions from it when, before you maybe got these, you were looking at some data points on a screen.
Speaker 1:Purchase the technology to have it in hand, but also to draw conclusions from it. Yeah, where, before you maybe got these these, you were looking at some data points on the screen. Maybe you got all this raw data, but you still had to analyze all of this. Yeah, and you also see there the maturity that this analysis is done and we we see, indeed, like there is no full support on this leg or whatever yeah, it's true and I think, but I think it think also for running it's.
Speaker 2:Yeah, I'm not sure because I think, like different sports have different degrees of subjectivity I think gymnastics has this part that is very subjective right and I think running is very objective right and I think it's very it can be very physics backed right which I think is very nice, I think also data. We also did a project on the changing the height of the saddle of your bike yeah, cycling geometry right, which I think is kind of. It's a bit relatable to this right like something that we analyzing.
Speaker 1:So how, how someone is sitting on a bike or how someone is running, the gait of someone is running exactly, and draw conclusions based on that, giving advice based on that and so there you just need a camera.
Speaker 2:I guess you don't need to have like special markers or all these things. You can kind of do all the yeah with good results.
Speaker 1:That's very interesting to see, like that, to uh, to get this to athletes where before, like, there had to be a very big expert team to get the same results. Um, and also, like, I think also the way the data is visualized today and you see that in, in, for example, the gate visualization, to really understand this is what I can improve. But also, uh, if you open the, the reddit link, I think that was a very interesting one like where you see the this one the 100 meter uh, sprint finals, um.
Speaker 1:And you see, uh, it's lyles that wins. But lyles actually starts very, very late. It's the person in the third lane, um from the bottom and he wins in the end. And after this we're looking at the people sprinting. After this there is a recap where you actually see per, you see the evolution of this based on numbers, and there's a way to visualize all this data and it makes it also much more engaging.
Speaker 1:So you actually see lies that he's really dragging behind yeah it's not first at all at the beginning lies that he's really dragging behind. Yeah, it's not first at all at the beginning, um, and then at some point, I think around 60 meters, like there is a this, this very big acceleration of him where the others are not really following, and then he ends up as first, and here we see the, the progression of that on the screen. But you also see like the a bit of the heat map, like where you see it here. Now the heat map is really like around the just before before the 70 meters. That's where Lyle actually won the 100 meter sprint Interesting.
Speaker 2:You can also give like feedback for a very targeted you know you should work on this.
Speaker 1:It's feedback for the runners, but it's also a way to get the audience at large more engaged into. What is this? What is actually happening here?
Speaker 2:That's true. That's really cool. It's really cool.
Speaker 1:Where, before this, this specific data visualization, you really had to be like an expert like you need to know, like, how does that 100 meter sprint work? Yeah, like you really need to to to see this, to understand a bit like what is happening.
Speaker 2:But well this makes it very tangible, for I think yeah indeed, I think it brings a lot of expertise to people that don't maybe immediately have it. Yeah, and I think the when you have, like when you understand how something is more complex, you appreciate it more. Yeah, right, and I think that that does that for sure, right? Also, one thing I just uh, in the beginning I know they're doing this really looks like a video game to me, the way that the images yeah, it's just me like. It really looks like.
Speaker 1:It's just like this seems like a bird's eye view of the 100 meters final yeah, so this looks like a game I had on the on the super nintendo back in the day, right it looks like that. I was like I'm really looking at players, yeah and and what I like about this, this data visualization as well. It is super simple yeah indeed like um visualization as well.
Speaker 2:It is super simple. Yeah, indeed like um, but it's. It's a good combination between like not complex, but like informative enough, yeah, but simple in a way that's like, yeah, we're not, we're not have, we don't have any audio, we're just just look at it and yeah, you know what. You understand it right. I think that's actually the best uh visualizations, I guess and, but it made me think also.
Speaker 1:It's like and also like this, so he helps article like how mature are different sports when it comes to? Analyzing data true, I think there's a big. There are big differences there, like we. We know that football like analyzing uh for key players, value of players, these type of things like. There's a long history of that, I think. At the same time, there are a lot of sports that are not taking the benefits that they should.
Speaker 2:Yeah.
Speaker 1:So I think, for example, um triathlon, um the triathlon they swam in the santa there was a bit of a drama there. No, it was a lot of drama. Let's not go into the drama, but there was a lot of uh discussions, uh, after the swim that there were a lot of currents in the santa. Really, apparently they're stronger, so they didn't know.
Speaker 1:You also saw it because they so I'm very close to uh to one side to avoid the, the currencies from the line, and I was thinking like whether or not they analyzed it before, because it's super easy to just use sensors and like to know like this is the best way yeah, like you take an rc boat, you put a current sensor under it and you just boat away over this to see like as an athlete, you need to take this path yeah, this is the best part current wise yeah and like it's, if you know the the best, best, best spot, current wise, it's probably gonna make a difference, big difference, a big difference, even yeah, this we're not talking about marginal gains no, no, no, you're talking about like and like if you, if you're in a sports where, like all of these olympics was like four years, you give everything of your life, yeah this is like.
Speaker 2:These are the things that you need to do.
Speaker 1:I'm wondering whether, whether the, the big names did this, for example I would be yeah, I mean, it's true.
Speaker 2:It's true. I think if they don't do that, they're leaving a lot of the people like a lot, a lot, a lot yeah, I think so as well, and and maybe also linked to triathlon.
Speaker 1:Again, there, there were a huge amount of falls Falls Like in the bike race, in the bike part, especially in the women's race. It was raining, so it was wet and probably don't quote me on this, but triathletes are not the best cyclists, probably. You're already here first, but this is a huge event.
Speaker 1:you work towards this for years yeah like you should up set up something like digital twins, like where you emulate, like you know how the course is going to look like. Yeah, you know what the conditions are that you that like the range of conditions you can expect. Like you, need to prepare for this right?
Speaker 1:yeah, I think I know I agree this is the culmination of like it's more than four years because, like you're probably into it, more than 10 years by then yeah, yeah, no, I I know, but I also think, at least in brazil, because I actually I I did a bit of judo actually oh yeah, oh what uh I did a bit judo um, and I think the, the, the sensei right, he actually I think he went, went, he went for competition like and he won gold medal.
Speaker 2:I don't know if it was Olympics or something, because he wasn't like fighting, it was more like the demonstrations of stuff and and he actually won the gold medal. But I remember, like he didn't have money to travel, he had to like kind of do some like raffle kind of money to travel. He had to like kind of do some uh, uh, like raffle kind of thing to collect money to go. Like the, the Brazilian Olympic delegation, whatever, didn't give him any money for anything right.
Speaker 2:So I think for a lot of these players, this place is athletes, at least coming from Brazil, like the, the. It's not like they. They have a lot of spotlight in that moment, but throughout the year they don't have almost any support, which is they don't have the means, and that's where maybe the quality part comes in.
Speaker 1:Exactly, yeah, it's.
Speaker 2:That's why it's important yeah, exactly, it's like you know, but I was really surprised to to hear about, like I mean, I also saw like interviews from other Olympic athletes and maybe the ones that get a lot of spotlight maybe they get a better facility, facility training, all these things. But I think for a lot of them there's a bit individual stuff. They have to kind of figure a lot of stuff out. The fact that they're there is already an achievement for them. So I don't know how it is for all the triathlon athletes and all these things, but yeah, I think for the ones that are big names, that they have the means. If you don't try, like you don't see, like what's the the floor conditions, you know, like do you need to put a tire, like rubber that is has a better grip, you know, especially for rains and all these things, you're definitely leaving a lot of the table and I think they're assuming you have the means exactly for it.
Speaker 1:You have to support for it.
Speaker 2:Like these are things like you need to understand the data yeah, and I think I think even with this Olympic AI agenda I think part of it they wanted to do workshops with people from the delegations on how data can be used to support them. So I think sometimes it's just knowing that these things exist and these things are accessible, right, like we saw for the running thing. Yeah, you just need a camera. You don't need specialized hardware. Really, you can do stuff on your browser. You just need a camera.
Speaker 1:You don't need specialized hardware really. You can do stuff on your browser, and that is because of where we are today Exactly the technology that exists today. Indeed, something else that I find interesting application that they're using apparently didn't know until I started looking it up is that they use AI to detect comments that are like a bit derogatory or inflammatory comments about athletes or aimed at athletes on social channels and to automatically flag these to the review or whatever, of this channel.
Speaker 2:About mental health of the athletes. I didn't want that, exactly, exactly, and I think that is very uh.
Speaker 1:It's a very nice uh initiative I thought so too.
Speaker 2:I thought so too.
Speaker 1:I I assume that they uh aligned this with the, the big social platforms to automate this to some extent, and that they that they have tested this I will hope so I hope so. Probably x. I said uh forget, I would hope so. I would hope so. Probably X has said fuck it.
Speaker 2:Not here. We don't do reviews, yeah, but that's the thing. For me, it really felt like Olympic is just a big company that have different fronts in which you can apply AI and they were just elaborating a lot of them Like this. What you mentioned is like content moderation.
Speaker 2:Yeah, it's very nice, you know so it's like different Nice that they're applying these things. But I also got a sense on when you look at all the things that they're applying AI and data analytics, for, yeah, it's just like stuff that you would see at companies, right, and it's like yeah, exactly exactly, but it became so mature that it has become easy to do. Exactly.
Speaker 1:Real away from the.
Speaker 2:This is an experiment, to where we're doing it, to like we're actually doing this data, indeed, indeed. So I think, in the end, for me it's like people are is ai going to be part of the olympics? I think it already is. It probably already has been for a while, right? The same way that ai is part of our daily life, right? Um, you had one more article, I think you wanted to bring it up. You had one more article, I think you wanted to bring it up. I did. Ai is trying to medal at the Olympic Paris well, it's the other.
Speaker 2:I think we already discussed some of those three ways AI is changing 2024 Olympics cool, maybe. I'll just put one last thing then. I thought it was also interesting again in light of the 1924 Olympics and the 2024 Olympics. So it's in 1924, football in color. So again, alibaba, ai Cloud. They were also sponsored and I also think this was a bit of an opportunity for some of these tech, like Intel had some stuff to really promote their AI services, but they basically went to 1924. So they have for, like, if you go to their YouTube channel, they have a whole bunch of sports and they actually went through the footage and they actually recolored using AI. So you see the difference here.
Speaker 2:So this is what I'm putting on the video for anyone following. You have the black and white footage and then now you have the colored one, right. So not sure how accurate it is. I'm not sure if you can be super accurate, right, because you don't have the true data, but it looks realistic, it looks very realistic it looks very and it's probably not very. It's not like far off right, it's probably.
Speaker 2:It's hard to know what it actually was right, exactly like you don't have, unless you had like pictures with color or paintings or something, but uh, but I thought it was cool. It brings again like a bit of a different uh, a different view, of refreshing way to look at something that is, that is quote unquote old right. So yeah, and maybe uh already thinking what's gonna be olympics in uh 21, 24?
Speaker 1:it's gonna be just robots, just just robots, yeah, just ai systems talking to each other yeah, exactly right, it's gonna be like one second. Olympic.
Speaker 1:Sometimes we're allowed to watch yeah, exactly cool yeah yeah, the only thing that I was a bit wondering also, like is about, and it comes to the judgment, to the, to the judgment system as well, like for gymnastics. But what triggered my mind actually was the, the athlete chat app. It was uh called athlete, the athlete 365, and you were already saying a bit like there's an lm that you at least can ask uh, uh questions about uh, but uh, practicalities during a race, like how do I prepare these type of things, where do I need to be, but also regulation questions, and I was wondering, like, how sure are you that there is no hallucination?
Speaker 2:yeah, sure, right? Yeah, it's like, yeah, it's tricky, unless it's just unless. Well, if you're really thinking, if it's not an lm, it's just the the vector search part. Yeah, right, question mark.
Speaker 1:Yeah but to be seen exactly, yeah, exactly. Because this is a bit like where who's at fault like I'm gonna try to jailbreak it, have a good prompt so that when I ask, is it okay if I use uh epo doping and but I've jailbreak the problem so, and actually this official app says yeah, it's okay, and I'm gonna say yeah, but but I had this uh dispensation like it's official ioc application yeah, it's true, I'm like, for this part, a lambs hallucination we need, we, we will have, let's, maybe, let's, maybe make a stem.
Speaker 1:We will have a solution next olympics for hallucination I think, so that I think I'm very confident. Yes, okay yeah, even if the solution is let's not use it okay, but we'll have a solution for hallucination next olympics, so oh actually I didn't take out this in four years.
Speaker 2:You can uh, you know, check with us here what's the solution four years, thank y'all thank you any big plans.
Speaker 1:You have tech it's summer, it's quiet time.
Speaker 2:I have some plans, too much to get into shouldn't you wear your? It's too late for now. A lot of code for me and usually it's slightly wrong. I'm reminded it's a rust here, I think the olympic rings should be more prominent this almost makes me happy that I didn't become I was like I'm really was embarrassed when I didn't form. Oh yeah, I went to the iphone speak to you today about large neural networks. It's really an honor to be here Rust, rust, rust, rust.
Speaker 1:Data Topics. Welcome to the Data Topics. Welcome to the Data Topics Podcast.
.png)
