Will Your Next Running Coach be Artificial Intelligence?

The future of coaching and how trail running presents a unique challenge for artificial intelligence.

Photo: Getty Images

Heading out the door? Read this article on the new Outside+ app available now on iOS devices for members! Download the app.

Look across the internet and you’ll find many different options for trail and ultrarunning coaches. In a May 2022 survey fielded by the American Trail Running Association, 14 percent of trail runners had coaches, and 4 percent were considering getting a coach.

With new Artificial Intelligence (AI) innovations, like OpenAI’s ChatGPT and Google’s newly-released Bard, will AI coaches compete with human coaches? How will these emerging technologies affect coaching as we know it? 

What is AI (Artificial Intelligence)?

If you ask ChatGPT to define itself, it will say: “the development of computer systems that can perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and language translation.”

The reason AI has become the tech world’s darling is that there have been incredible advances in a type of AI known as “Large Language Models,” or LLMs. These models use natural language processing and have the capacity to quickly complete complex tasks. The company OpenAI is leading the way with ChatGPT, which can do things like summarize a long body of text, infer meaning, transform data from one language or format to another, and even expand text: you can literally give the model one sentence and ask it to complete your paragraph.

According to Stanford University’s Human-Centered Artificial Intelligence group, the new models are incredibly powerful: “These models have already been used to co-author Economist articles and award-winning essays, co-create screenplays, and co-construct testimonies before the U.S. Senate.” 

How AI is Being Used in Coaching Today

A December 2022 report by McKinsey & Company found that AI adoption by companies has more than doubled globally since 2017, with 50 percent of companies stating that they’ve adopted AI in some form. 

Currently, we have yet to see an app that’s fully leveraging the new technology to coach trail runners. Instead, what you’ll find is a growing marketplace of digital training tools that use existing, more stable forms of AI, like machine learning. These apps can analyze your data and dynamically adjust your plan, based on pace, heart rate, and in some cases, power meter data. Companies leading the way include TrainAsONE, AIEndurance, Athletica, HumanGo, and enduco

RELATED: AI and a Brave New Running World

While these apps aren’t designed specifically for trail runners, they’re designed for endurance training more generally, and promise to provide a training plan that’s tailored for your race date and your training preferences. TrainAsONE, AIEndurance, and Athletica all confirmed that they are investigating ways to integrate new AI like ChatGPT, and it won’t be long before we start seeing more uses of the technology.

Athletica initially created an app that would compete with human coaches, before recognizing that AI and human coaches could be symbiotic. Co-founder Paul Laursen was the Performance Physiologist for High Performance Sport New Zealand for the London and Rio Olympics before he started the company.

“When we first introduced Athletica, alarm bells went off for coaches,” he says. “All my life I’ve been helping coaches, and now I go to make Athletica and it’s almost like it’s a competitor for coaches.” 

The company pivoted, deciding to create a version of its product for coaches that’s designed for coaches to use with their athletes. Coaches can use it to “automate” the creation of training plans using Athletica’s training methodology, but still maintain the one-on-one interactions with their athletes. The company sees a future where its AI app is used as a tool for coaches, rather than creating a product that would replace them.

Will AI Replace Trail Running Coaches?

Cliff Pittman is a trail and ultra running coach and the Director of Coaching at CTS, an endurance coaching company that’s been around for more than 20 years. The company has many well-known ultrarunners as coaches on staff such as Jason Koop, Stephanie Howe, and Andy Jones-Wilkins. 

While Pittman hasn’t tried any AI tools yet himself, he thinks that AI may be able to help with the process of creating training plans for athletes, and possibly even something dynamic, that can adapt based on an athlete’s training data. In the near term, he sees AI helping to serve runners who weren’t using coaches previously, whether due to finances or other reasons. 

RELATED: Humor: The Too-Smart Smartwatch?

Pittman recognizes that AI has limitations, but he thinks that the technology may evolve to disrupt certain types of trail running coaching. In the long term, he envisions AI replacing coaches who handle a large volume of athletes, with less focus on personalization and less emphasis on facetime. 

“I think that AI is going to be able to provide that exact same service at a lower cost,” he says. “And if athletes already don’t have individual connections with their coach, if they’re not having phone calls with their coach, if they’re not exchanging emails and text messages back and forth, they might as well just go with AI.” 

It remains to be seen whether AI can produce the level of personalization that human trail running coaches include in athletes’ plans. Will AI tools be able to know that an injury-prone athlete may respond much better to doing their strides and workouts uphill, or even on the bike, for lower impact? Will it be able to nurture and guide a postpartum athlete who is building back her training volume and intensity?

As a professional trail runner and coach, Cat Bradley coaches 25 athletes through Boundless Endurance. Her team of co-coaches recently experimented with an AI tool, as they were wondering about how it would affect their business. “We used it, not to make a trail running specific plan, but just a running plan,” says Bradley. “First we started with 5K and went all the way up to 100-mile. It spat out plans that made sense; it was kind of crazy.”

RELATED: WATCH: The Canyon With Cat Bradley

At the time, Boundless was considering making individual training plans for purchase. The results of the test were so compelling that Boundless pivoted away from that idea. Instead, they’re focusing their energy now on the human side of their business, which is one-on-one coaching. “If someone wants a plan, they can get it [with AI]. We shouldn’t invest too much time and energy into this right now,” she says.

Emerging Possibilities of AI in Athletics

Palo Galko is a trail runner, triathlete, and IT Manager who lives in Melbourne, Australia. He’s been programming since childhood and loves developing his own software programs to help him analyze his training data. 

On April 6, Galko created an experimental athletic analysis tool using AI, which he posted on his Twitter feed at 1 A.M. “It was late. I finished, put it on Twitter, and went to sleep,” says Galko. “I woke up in the morning, and I’m looking at my phone, I see 22,000 views!”

His open-source tool, BambooAI, uses new AI to create a super-powered training analysis tool: you can input your Strava or Garmin FIT file, and use human words to analyze the file. Galko also added a feature where you can provide the program with PDFs that are relevant to your analysis, such as a physiology textbook, or a new research paper on a particular training method. The model will essentially learn the contents of that PDF, and you can then ask questions about your FIT file in relation to concepts in the PDF. 

RELATED: Five Key Takeaways from Camp Strava 2023  

For example, you could upload the activity file from your most recent 100 miler, as well as a PDF about pacing strategies. With that information, you could ask questions about how well you paced in the race, like whether you excitedly ran too fast in the beginning, or on climbs in the first half. It sounds promising, but Galko is the first to admit that his tool is experimental, and occasionally gives misleading answers.

Galko believes that the power of the new AI is that you no longer need to be a major data nerd to do athletic analysis. “Previously, you would have to be a data analyst or programmer to do that, or rely on the tools that the likes of Garmin or Strava provide to you. But now, really, you can talk to your data in the natural language, and that’s the beauty of it,” he says. “You don’t need any technical skills. You don’t need prior training to get some insights from your data.”

ChatGPT entered the market six months ago and it’s likely there are many curious athletes, scientists, and entrepreneurs creating all kinds of new AI training tools in their basements. Galko shares, “In my 20-odd years in the IT industry, I’ve never seen anything even remotely close to what’s happening now. It’s nuts.”

OK, How About AI as an Assistant Coach?

With smart trail runners like Galko working on inventing AI tools for training, it’s just a matter of time before AI can assist coaches with tasks that machines do well, such as creating training plans, and doing data analysis on years of data in athletes’ training journals.

Laursen thinks that AI may become better than humans at certain aspects of the coaching process, like making training plans, and adjusting to any unexpected changes in athletes’ training loads. He thinks humans will continue to play an important role, as there is so much more to coaching than the training plan. “There’s almost an infinite number of other psychological, cognitive, emotional sort of stuff that coaches are still going to be better (currently at least), than a machine,” says Laursen.

RELATED: Why Elite Athletes Are Getting Serious About Mental Coaching

One-size-fits-all training plans may die as AI training apps like Athletica emerge, and there may be a day when coaches use an AI sidekick to enhance parts of the process, like training plan development, or training analysis, to systematically detect athletes’ fatigue, injury risk, and response to training. Perhaps the business for off-the-shelf training plans may dwindle, but athletes who want to build a long-term coach-athlete relationship will likely stick around.

Trail Running Presents Unique Challenge for AI

As app inventors rush to make AI-based training apps, there is an elephant in the room: will trail runners be receptive to such a high tech approach to training?

Pittman thinks AI may have a harder time measuring training stress for trail running compared to other endurance sports like cycling or road running, as the trail itself creates a new variable that’s simply hard to measure and compare.

“It has to be able to measure that [training stress] in some way, whether that’s duration or intensity,” says Pittman. “Certainly that can be challenging when there’s variability in place, such as running up mountains, running on a variety of different surfaces. I think that’s where technology has yet to really be able to measure training stress, is the variability of surfaces.” For Pittman, this is where the human element shines, as human coaches can have frequent conversations with athletes to discuss their training, and seek-out an understanding of those more subjective variables. In the meantime, he thinks that AI may be more readily useful for cycling, track, and road running, which are sports that have fewer variables, and a smaller margin for error.

Trail and ultra running also spans so much more than just the running part. Galko says, “Running downhill on muddy terrain, it takes skill. You cannot just turn up, and do it. So there is a very large technique component there as well that AI might not be able to help.” 

The Human Side of Coaching Isn’t Going Anywhere

The common thread is that everyone agrees on the magic of real, human coach-athlete relationships. AI may be able to bang out fairly strong training plans that can serve as a decent starting point, but the training plan is only one reason why an athlete would hire a coach. 

So, what is it about the human experience that’s so special? 

For Bradley, it’s this human role that differentiates coaches from AI alternatives. “The personal support of having someone to talk through fears, talk through nutrition, and really individualize—this is the stuff that AI can’t do, and won’t be able to do,” she says. “That connection, instilling confidence, the experience that we have as coaches, and really personalizing the training.”

RELATED: Trail Runner Coaching Concierge

Still, coaches shouldn’t be totally dismissive, as AI has been shown to provide meaningful emotions in a coach capacity for other industries. Recent research is finding that chatbots can provide motivation, and even empathy. 

In a 2022 study published in the Journal of Medicine Internal Research, researchers studied the use of chatbots in a smoking cessation program. They found that the chatbot did lead to feelings of accountability and engagement, stating that “users described feeling that Quit Coach [the chatbot-powered app] cared about them, which helped foster a relational bond and promoted a feeling of accountability to Quit Coach to succeed.”

As an athlete, would you feel as motivated for a big week of training if you knew it was just a machine reading your log and giving you feedback, rather than a human who you know and trust?

For Pittman, the answer is no. He says that many trail runners will hire a coach for the accountability.

“I’m coached by Jason Koop,” he says. “And, you know, whenever I’m racing, I want to make sure that I represent him well, too. When you have that good coach-athlete relationship there are heartstrings involved. You want to perform to represent them well, to make them proud of you.” He doesn’t think this is something that AI can do.

Bradley thinks that a key differentiator is that she’s experienced most of the common ultra running issues herself over the years, and she can help athletes prepare. “The AI program won’t be able to tell you what mile 65 feels like or what signs to look out for,” she says.

The Future is Here

The future likely has some combination of AI and human coaching, to get the best of both worlds. 

According to Pittman, it’s important for coaches to stay open-minded to embracing technology, or risk being left behind: “I can’t help but think about major brands from my childhood, such as Blockbuster and Toys-R-Us,” he says. “These giant companies refused to embrace technology, as elementary as it was back then, and those companies are no longer around.”

One thing is clear: coaches should keep focusing on the human part of their business, because in a sport that challenges its participants to the limit, that’s not going anywhere soon. Who knows, maybe in the future someone will develop an app for 100 milers that mimics coaches’ voices and personalities for a fun and entertaining virtual pacer, finally a way for coaches to be at every athlete’s race at once! We’ll have to wait and see. 

Alicia Woodside is a trail running coach, runner, and product manager based in Squamish, Canada. She offers coaching for trail runners through Woodside Adventures. She did not use AI to write this article.



Want to Know What It Takes to Finish at Western States? Just Ask Hellah Sidibe.

Find out what happened when this six-year run streaker and HOKA Global Athlete Ambassador took on an iconic ultramarathon in California's Sierra Nevada