Why fitness and lifestyle coaching is a great fit for AI
So you may have heard the “AI for X” pitch many times already, and you may be wondering exactly why it makes sense to marry fitness coaching and artificial intelligence.
This is part of the drill for emerging technologies. The hype has been particularly strong around AI, which is no surprise considering the techniques that make up the field are very general-purpose and can be applied to almost any domain, as long as there is data.
This does not mean that as long as there is data, one should always apply these techniques. Not everything is a good use case for AI. But fitness and lifestyle coaching truly is, because — as we will explore below — the profession has a lot of characteristics and elements that allow machine learning algorithms to play to all of their strengths.
This is the second part in a series of four blog posts where we explore how AI could impact the fitness and lifestyle industry, mitigate the steady increase in lifestyle-related health problems and make our lives better.
The first part made the case for lifestyle coaching as a way to address this. In this post, we explain exactly why there is reason to believe AI will impact the field of coaching.
An exercise in prediction-making
To understand why AI is immediately applicable for many aspects of fitness and lifestyle coaching, we should look briefly at how our current idea of AI actually works, and when it can be useful.
AI is well suited for helping out with tasks that require just the right amount of human reasoning, in settings where the most important information required to solve a task can be made available for an algorithm to make predictions.
The complexity of the task has implications for the amount of data that needs to be gathered to train an algorithm to perform it. This is what we call machine learning (ML), which is at the heart of most of what we today refer to as AI.
Humans — like AI — make predictions all the time. The daily life of a fitness and lifestyle coach is positively brimming with prediction-making. When a coach advises a client to do Z (like eating a single slice of cake at the party instead of abstaining completely), he or she is predicting that Z will somehow contribute to a positive outcome Y (like greater and more sustainable weight loss in the long term) for the client.
To make this prediction, the coach has likely made use of information X about the client, their history, and the current context. For another client with different attributes and goals, the advice might be completely different, because the coach has learned the relationships between the variables in the equation and is able to adapt his or her prediction to this new setting.
Learning from data
While current approaches to AI are far removed from any kind of human intelligence — such as that possessed by a coach — the above is still a decent example of how machine learning algorithms work on the conceptual level. They learn relationships between inputs and outputs from historical data and use this to make predictions about the future.
In the above example, a machine learning algorithm trained on plenty of examples of information X, behavior Z, and outcome Y would learn some approximation of how these things relate to one another and could be used to make the same prediction as a coach at a fraction of the cost and at near-infinite scale. The data is the key to unlock this potential.
Traveling back to the present and putting our feet on solid ground again, it is obviously worth mentioning that there is a lot more to coaching than prediction-making. Coaches instruct, demonstrate, motivate, and engage in no shortage of casual banter with clients.
Sure, you can scale instructions and demonstrations using video, and sure, AI can yell at you through your headset — and you can even have a reasonably coherent conversation with it at this point.
Does it feel the same as actually having a real-life coach? Probably not. Will we get there? Maybe. But the most interesting question right now is: can AI do a lot of the other things, leaving coaches to focus on what humans still do best? Most certainly!
So while replacing the full personal coaching experience with a robot — even just a purely digital one — is likely not going to happen any time soon, the profession of coaching is ripe with small tasks where AI can both amplify human coaches to make even better decisions, and replace them entirely by interfacing directly with consumers so humans can spend time on what humans still excel at: complex reasoning and providing that personal touch.
In the next blog post, we will look at concrete examples of what AI systems for coaching could look like for coaches and clients and what kind of tasks they might perform.
