Speaker 1: 0:00

So you're thinking about running, but not sure how to take the first step. My name is Brian Patterson and I'm here to help. Welcome to Brian's Rompod. Well, welcome back to Brian's Rumpod. And first of all, I'm recording this on New Year's Day, 2025. So just wanted to say and wish you all a very happy new year and you're getting over that festive break and I hope you all had a great time. So, to wherever you are in the world and we do get listened to all over the world and I just want to say happy new year and I hope you I can continue to provide you with some great content for 2025.

Speaker 1: 1:08

Now, in today's episode, we are going to be looking at a very topical subject and that's artificial intelligence, because I know it's very much part of what we do in our day-to-day lives and also it is very much part of what we do in exercise as well, and running is no stranger to it. So well, what we're going to be looking at, looking over the show notes, so we're going to be looking at what is AI, what is machine learning be looking at is what is ai, um, what is machine learning? Some of the applications that have been used uh for uh, um, that incorporate ai, maybe some of the drawbacks of ai. Um, how, maybe a human element is maybe better than um, than ai, and then I'll kind of give you maybe what's the future hold for AI and running. So maybe going over a few of the same things as well. Anyway, so what is AI?

Speaker 1: 2:18

Well, the basis of artificial intelligence is a concept called machine learning. Now, machine learning is the process from which a computer is able to learn new information independently, that is, without being explicitly programmed or given that information. It uses something called a neural network, an artificial replica of mechanisms in which the human brain learns to analyze large amounts of data and derive a summary of the information based on patterns it has detected. So how does this? You know you may be asking how does this affect running? Well, what types of AI could be used in running? Well? Natural language processing, image recognition and predictive analysis. So hopefully I'll kind of go into and explain what those mean. So there are a number of ways that computers can use machine learning, but the most popular are natural language processing. So then there's image recognition and predictive analysis. So natural language processing is a type of AI that is capable of producing text based on its programming, plus information it finds on the internet and from human input. It uses machine learning to learn how humans communicate and will adapt its responses to give answers that meet specific criteria. Computers can identify and analyze the visual information using image recognition as well.

Speaker 1: 3:47

Now there are AI tools out there which many, many of you may be familiar with. So there's ChatGPT and then there's DALE. Now one of the most popular tools is a natural language model called chat gpt and there's the image creator, which is called dal, e now ai in terms of in current applications. Um, I have, uh, like, an example here which is really fascinating example. Ai is very much going into this sci-fi idea to something runners are actually using. So let's jump into this.

Speaker 1: 4:26

Who ai the ai coach thing? There's this guy called renato um, and I will leave a link in the show notes to this article. He wrote an article on medium about how he built his own ai coach. It's an amazing story because it shows how he can, using expert knowledge is available to anyone. His story is something that all of us can relate to. He wanted to personalize his coaching, but he couldn't afford it, so he even used a tool called TrainAZone, which is an adaptive training platform. Basically, it uses an algorithm to adjust your training based on what you are doing. He felt he still felt something was missing, so he goes ahead and creates his own ai coach, called the 80 20 ultra coach. As I said, I'll include a link in the show notes. It's designed to give runners advice on everything training, what to eat, race strategy and so on. Plus, he made it totally free.

Speaker 1: 5:28

Now, no AI can replace a human coach. So, for starters, you don't have the human relationship for someone encouraging you while you're training. However, ai, in this case, is great for giving you data-driven feedback consistently. Now, for some people have asked a general AI tool to create a plan for them. Sometimes it can be a bit hit and miss. One guy did this and found that it spat out a plan for a 50k race when he wanted one for 5k race. Maybe he got confused with the lack of zero. So it's important to search for an AI tool that is built for running. So maybe give this 8020 UltraCoach a try. I know I will in the new year.

Speaker 1: 6:11

It can also be used for running form analysis. Now there are other uses in this space, but AI can be used to help analyze our running form. And there's this application called Ochii again, I'll put a link to it in the show notes which analyzes your running form. Now it says on their website transform your running with our advanced ai and biomechanics algorithm. Simply record a video with your smartphone for a full body analysis. Receive personal exercises to enhance your technique, efficiency and performance.

Speaker 1: 6:46

Developed by former high level athletes, ochi provides insights into your strengths and weaknesses and help you prevent injuries and improve your running form, all without any extra equipment. Now the running form analysis the app uses advanced AI and biomechanics algorithms to analyze the video on the running, providing a full body analysis. Then it gives you personalized exercise plans based on the analysis, and the app provides the tailored exercise plans to adjust weaknesses and unhearthed strengths in your biomechanics. And these exercises aim to improve running efficiency, build strength and prevent injuries. Continuous improvement Well, the app provides regular updates and progress tracking, offering new exercises that adapt to your improving form, ensuring sustained development Expertise. Ochi was developed by former high-level athletes, including a physical therapist with biomechanics lab experience, ensuring the app is both practical and based on scientific research.

Speaker 1: 7:53

Convenience Runners can conveniently record a video using their own smartphones for analysis. They can also share their progress analysis with coaches or peers. Accessibility the app does not require extra equipment. It also is available in both iOS and Android. Now, I think how it works is that you point your phone or maybe get someone else to record your run, so you could either put it on a stand and hit the video record button, and I think maybe you do this through the app. I'll have to find out a little bit more about this, but, like sort of other apps that do sort of image recognition and analysis, I'm assuming that this would be the same way. Then it uses a set of, as I said, advanced AI and algorithms to look at your running technique and give a guidance on how this can be the same way. Then it uses a set of, as I said, advanced AI and algorithms to look at your running technique and give a guidance on how this can be improved.

Speaker 1: 8:48

Now, the thing I worry about with this is that maybe you're trying to shoehorn technique into something that is very natural to you. So, even when I had the running analysis at Move Physio some time ago and I think I did do a podcast about it and I if, if you go back into our archive, um, I will. Maybe what I'll do is I'll give you a link in the show notes to it. Um, I had aaron, who was the physio slash personal trainer, doing basically a video of how I was running and then he was sort of giving me feedback about how maybe I was over striding and I was really more of a heel to toe strike rather than sort of midfoot strike. So that is something that you have to try to. It's going to take time, it's not going to be over. You know, over a week you know you suddenly get this perfect technique. But I think what it does is that it will give you exercises to help strengthen your body so you can adapt to a better running technique, and I think you have to just think of it as like very much a process of improving your running technique. And then I think maybe you take a video at one stage and then at another stage, maybe in two months later on, maybe eight weeks later, you do another video and then you see what the feedback's like and then it will probably give you a few pointers whether you're improving or not improving, but hopefully you will be improving because you're doing the exercises as well. It may be worth, you know doing this in conjunction with a coach, and they can and they can help as well.

Speaker 1: 10:47

I must admit, as I said, I haven't tried it. However, running is very much an individual thing. As I said, it's not like we're training for the Olympics or being a professional athlete. Most of us are just doing park run or doing a 5K or just trying to get on. You know, get the treadmill in the gym, including the warm-up. We just want to get fitter and enjoy our running. So getting stressed about it if our forefoot is not landing well or head position isn't in the right place might not be the right way to go. So it's just a tool to help you with your running. In saying that, that, I did say it does recommend exercises that can help with your running form. As I've said before, in this it's not just about the running, but also about doing strength work and exercises specific to running as well. On the top of my head, I can think of two Olympic athletes who had very odd running techniques Paula Radcliffe, the GB, great Britain marathon runner, who was also very successful in the 10K with a head moving. You know, a very distinctive running style. And then there was Emil Zatopek, who was one of the most successful 5,000, 10,000-meter and marathon runners in Olympic history, who won a total of five Olympic goals from 1948 to 1952.

Speaker 1: 12:12

Now performance analysis insights how does AI help with performance insights? Well, strava's new athlete intelligence tool uses artificial intelligence to help runners understand and improve their performance, and so here's how it works. It does an analysis of workout data AI analyzes the complex workout data to provide personalized insights. This helps runners understand their recent running performance. The tool analyzes data for the past 30 days. Identification of patterns the athlete intelligence identifies as patterns and peaks an individual training to give a clear picture of progress, and this helps runners see how they are progressing towards their goals. Specific metrics the tool provides insights on running pace, power, time spent in specific heart rate zones. Relative effort, which is Strava's metric for gauging intensity. Personalised training tips the tool offers tailored training tips to help runners improve their feedback endurance and stamina, and speed and stamina. Motivational feedback the athlete intelligence celebrates personal bests and other training milestones to keep the runners motivated.

Speaker 1: 13:25

Accessibility the feature is available to Strava subscribers through their smartwatches or mobile phones and it is available in public beta mode in 14 languages. So what you can do if you do use another app to record your running, then what you can do is you can use other apps which will upload it to Strava. So even if you aren't a I mean I know I am a subscriber on Strava but if you do regularly use like I use RunKeeper to record my run, my run because run keepers great at doing uh, setting up interval sessions. Or, alternatively, I use um because I use an apple watch, I can use the apple watch app on the, on the watch again to record my run and, again, that's really good at doing interval sessions and I can upload that to Strava and it can give me feedback AI feedback as to my runs. I will give you a link to my Strava in the show notes.

Speaker 1: 14:37

This is very easy to interpret the insights. The tool provides an in-depth but easy to interpret insights to effort and activity trends. Continuous improvement Strava says that the tool will continue to evolve over time. So and I think that's another key thing here is that, although AI is is very much kind of in its infancy, in some ways is for relying on historical data and over time though it may be in some ways beta and it's kind of maybe hit and miss with either training plans or looking up and giving encouragement on various runs that you've done. It may be a little bit hit and miss here and there, but over time the idea of natural language processing or machine learning, as the name suggests it will get better and therefore it will help sort of become more honed in to what your needs are. So in in summary, strava's elite intelligence tool uses AI to provide runners with personalized, comprehensible insights based on their activity data, and this helps runners understand their progress, adjusting their training and stay motivated. And, as I said, I'll share my running stats in the show notes.

Speaker 1: 16:06

Now what's the human element sort of AI versus coaches? Well, there's the lack of best practices. Even the best AI-generated training plans may include suggestions that are not considered best practices and lacks evidence-based planning, the generic approach. Ai may not always individualize plans based on the runner's specific needs, health, training history and age, as a human coach would. Then there's a limited understanding. Ai may not fully grasp the nuances of the runner's situation or be able to ask follow-up questions to refine a training plan. For instance, recent studies showed that the quality of AI-generated training plans depended heavily on the detail of the input provided by the runner. If the runners lack knowledge or detail, the plan would be of lower quality. Ai may not ask the clarifying questions like a human coach would, leading to less tailored plans, the inability to adjust to individual needs.

Speaker 1: 17:13

A human coach considers many factors when creating a plan, including the runner's current fitness level, health training history, age, the demands of the race and how long there is to train. And, as I've said, we've talked to quite a few coaches on this podcast and I know some of the coaches like that one-to-one relationship with their particular client, which means basically, they're able to get a feel for what their needs are. Do they like running? I know it's an obvious question. Do they like running? What does running mean to them? How do they feel when they're going out running? What do they think about the last session, the times of day they like running? I know if I had my choice, I'd be going out running early in the morning because I have more energy at that time. So there's lots of questions that a coach would ask. Ai may not account for all these factors or in a comprehensive way. No human element. Ai cannot replicate the communication, trust and support that a human coach provides.

Speaker 1: 18:23

Human coaches offer personalized guidance, mental preparation, race plans and ongoing accountability that AI cannot match. Plus, you're interacting with a human being. I know it's an obvious thing, but you're interacting with someone and they're able, and I feel that having someone else there with you and I feel that having someone else there with you where doing your running, I think is great. So I know sometimes running can be quite a lonely sport, although sometimes people might like that. You know, maybe you just put your headphones on, listen to music, podcasts or an audio book and then off you go, and then maybe the training plan you're doing may give you some feedback on what kind of sessions you should be doing, like, for instance, I know for the Apple Watch sessions they can create.

Speaker 1: 19:10

You can say, well, go between this particular speed, you know, either a six-minute or six-minute kilometer to 530 kilometer for 400 meters, that kind of thing. But if you've there someone there with you and saying, um, I know I fly from one of one of the coaches I think it was andrew wren who who said, yes, you go from uh, this uh tree to the next tree, you do uh, high repetition in terms of your have, a relaxed lip, relaxed mouth, you're doing a high turnover with your legs and from the next tree, whatever. So you've got someone there with you who's telling you all these different things. So I know, in time maybe AI will kind of be able to do that, but then it's just it's not someone there with you, anyway. That's why I thought of it that, but then it's just it's not someone there with you anyway. That's that's why I thought of it.

Speaker 1: 20:06

So, um, the limitations of current ai models. So a study uses a generic ai model, chat gpt3, to generate trembling plans but found they were not optimal. So but the more detailed the input, the better the resulting plan. As I've said before, however, even with a more detailed input, these plans may not be in a par with those of an experienced coach. So there's potential for poorer advice. If a runner provides limited information to an ar program, the training plan plan generated could be poor, following the principle of garbage in, garbage out. So it's really important that the information I suppose that's the same, obviously with a a coach as well that you're providing them as much information so they can tailor a plan which is more that meets your needs from a coach's point of view. If they're quite an experienced coach, they know with all their experience and they're going into their data bank experience they have that they can give you a plan that's more suited to you. This is a problem because many novice runners may not know what information to include, and the ai is unlikely to ask those clarifying questions in. In conclusion, while AI can serve as a foundation for training, it currently lacks the depth of comprehension and personalization approach of a human coach, especially when it comes to evidence-based planning.

Speaker 1: 21:35

Consider the scenario of hiring a running coach You're not just acquiring expertise, you're also, in my opinion, gaining the human interaction I said, and unfortunately, ai cannot replicate that. Moreover, I firmly believe that, from a mental health and encouragement perspective, a human coach is the ideal solution. That's why people opt for personal trainers. I recall when I was a personal trainer, I generally felt that my clients did not want to let me down and consequently, they pushed themselves to go the extra mile. So what's the future of AI in running? Now? We can agree that the cat is out of the box. There is no way we can put this back. As more and more data is uploaded to the web and that we become more sophisticated in utilizing that data, we can agree that there is no substitute for having a real human coach. So what does the future hold? Well, we can look at more personalized training plans.

Speaker 1: 22:31

Ai can analyze a runner's data, such as activity, health stats and feedback, to create customized training plans. This can help runners optimize their training, avoid injury and achieve their goals. Ai-powered tools like 8020, ultracoach, strava's athlete intelligence, offer personalized recommendations and training tips, and this the quality of these plans will depend on the level of detail provided by the runner, with more comprehensive input leading to better results. So these things have just come out in the past year. So I know Strava didn't have the athlete intelligence beginning of 2024. So this has just come on in the last few months.

Speaker 1: 23:14

Biomechanical analysis AI, through computer vision, can provide real-time analysis of runs and movements, and this can help identify areas of improvement, prevent injuries, enhance running efficiency for instance, using odd cheek and help with advice on running technique. Real-time feedback what about smart treadmills with cameras and sensors could give runners and coaches a comprehensive view of the athlete's movement, psychological factors, strengths, areas, term for improvement. This real-time feedback could allow for adjustments to training plans and running form on the fly. Advanced data analysis AI can analyze large amounts of workout data to identify patterns, peaks, trends that may not be obvious to the human, and this could help runners understand their progress, identify areas of weakness and make more informed decisions about their training. Stravis athlete intelligence an example of this, providing insights on pace, power, heart rate zones and relative effort.

Speaker 1: 24:18

And the more you put upload, I know. I had a look the other day and I was saying, well, has this done? Looked at the run that I did in 2023, a long run that I did and yes, it did. It actually gave me some feedback about that run, some AI data feedback on that particular run. Race strategy this is probably more for more serious runners. Ai can analyze data from past races to help runners develop a personalized race strategy. This may include pace recommendations, nutrition plans and other advice to optimize performance for race day.

Speaker 1: 24:54

Shoe design Now, this would be brilliant. Now AI can be used to design running shoes that are specifically tailored to runner's needs, such as comfort, support and durability. However, as I said, there are limitations Dependence on input. Ai generator training plans are heavily dependent on the quality of input provided by the runner, so you can't just say just give me a 5K training plan. I know for some apps out there I know for RunKeeper they can give me a 5k training plan, but I think the more information you provide like when you'd like to have when, when do you run, or when was the last time you did a 10k run or a 5k run, that kind of thing the more more information you provide, then the much more detailed and better the plan will become. Try not to be fooled into thinking that this will be the answer to your prayers into getting the PB, and I think we'll maybe talk about this a little bit more in the summary. I realize we're just scratching the surface now and, as I said, the modeling and algorithms will improve. The recommendations will get better Now, in conclusion, now, without doubt, we're witnessing a watershed moment in technology.

Speaker 1: 26:13

For instance, there was the internet, and then we had Web 1. I don't want to get too techie, but I thought I'd give you a little bit of a history lesson. I don't want to get too techie, but I thought I'd give you a little bit of a history lesson. So, as I said, there was Web 1, which was very basic websites, static websites or landing pages. Then along came Web 2. And this meant connecting people with the rise, dare I say it, of social media. There was interaction on websites, the use of what's called a software language, called javascript. So before there was what was called html, which was kind of the the bones or the structure of the web page. Then you had what was called css, which basically has all the coloring to the web page, and then you had what's called javascript and that basically you know if you press this button, it is this and that, or sliders and that kind of thing, so you use JavaScript for that. But then you had sort of like backends, so you had access, you had mobile access, you had websites becoming more responsive to screen size and therefore becoming more dynamic, and e-commerce, so you're able to buy things like you know, amazon website or other e-commerce websites, like buying your shopping online. Then we've got Web3, and this is about NFTs, blockchain, cryptocurrencies, which I don't really understand VR, ai, um, and added to this there was ai. So you know that's. That's part of word three.

Speaker 1: 27:51

So we, as runners, have always been looking for the edge and also for our next gauge, and we are no different, different than others in this space. So since the first time I bought an iPod Touch and used it to log my runs, I'd be fascinated about running technology. How fast was I going? How far was I going? I remember when I used to live not very far from here and I used to go out running, I used to map out the distance of where I was running, of, using, uh, the odometer on my bicycle, my bike. So I go and, let's say, do a map out of 5k and then the next day I'd run that 5k there and back and I've no 10 done. No, I've done 10k sort of thing, whereas now you get the real-time data with satellites or um telling you speed, calories, how far you're going, and all that kind of thing.

Speaker 1: 28:55

So ai tool, um is just a tool. Ai is just a tool that can give us guidance and, at the end of the day, it can't do the running for you. And I think, as I said, if we're just looking to improve, enjoy running more, make it more enjoyable, and now that you're given the data, this is right down our street. However, as I've discussed, this does not replace the human interaction a coach can give you. So I say it's a fantastic time for us as runners, in terms of the technology and what we have in our wrist or what we have on our phones, in terms of utilizing that data and processing that data and maybe giving you some form of training plan or advice going forward.

Speaker 1: 29:52

But let's not forget a human element, I'd say the coaches out there who are also recognizing as to what your particular needs are. So AI is great and hopefully maybe I'll explain it to you in some way as to what it's all about and I'm no expert, but this is my take on it but at the same time, let's just use it as a tool. I certainly like using it and it's great seeing all this information and all this feedback, especially from the Strava Athlete Intelligence. But then again, if you cannot afford to get a coach who can give you some guidance, then great, go for it. Thanks very much to listening to the podcast and looking forward to speaking to you very soon.