Is AI Fitness a False Proposition When Even Nationwide Home Confinement Can't Boost It?

[baoshi.rao]

Exercising at home has become the best solution to combat weight gain during this period. However, previously hyped features like 'smart fitness' and 'AI personal trainers' seem to have fallen short. Is AI, which even nationwide home confinement couldn't popularize, a false proposition for the fitness industry?

From the initial 'lying down to contribute to the nation' to the nationwide fitness movement, home workouts have become the best solution to avoid gaining 'ten pounds at home.'

Especially as sports bureaus in cities like Beijing, Shanghai, and Harbin issued notices advocating home fitness during the pandemic, it suddenly felt like exercising was a 'mission ordained by the state'—how motivating!

But unlike its shining performance in the medical field, AI's contribution to home fitness has been lackluster.

Following live streams and workout videos has become the most popular exercise method during the pandemic. Fitness coaches and Olympic athletes have turned into livestreamers, and the Nintendo Switch Ring Fit Adventure's price has skyrocketed, becoming the best investment of the year... Meanwhile, previously hyped 'smart fitness' and 'AI personal trainers' seem to have played little role. Is AI, which even nationwide home confinement couldn't popularize, a false proposition for the fitness industry?

The COVID-19 pandemic has not only made people realize the importance of immunity but also forced the fitness industry into a frenzied online migration. Home, online, exercise—each term seems to point directly to AI fitness. Yet today, people would rather follow square dancing in makeshift hospitals than work out with an AI trainer.

Why has AI suffered such a major setback? The answer lies in its previous application scenarios.

Since the AI wave began in 2017, its introduction to the fitness industry has caused quite a stir. Broadly speaking, it has focused on three scenarios:

1. Replacing Human Coaches in Gyms: Beyond rent, the biggest cost for gyms is human trainers. With landlords holding the upper hand, AI was seen as a quick solution to replace human coaches—a theoretically feasible idea.

For example, in 2017, Du Wuping, the developer of Tmall Magic Box, launched a holographic interactive training system called Smart Wall. Using touch-sensitive wall and floor screens, it monitors heart rate, tracks movements, provides AI scoring, and enables interactive workouts.

Similar AI-powered gym projects emerged, like VENTO in Beijing's Wangjing area, which uses AI devices to test users' jumping ability, agility, and muscle flexibility. The data is then synced to smart wristbands and connected to fitness equipment for automated adjustments during workouts.

In short, by upgrading fitness equipment and spaces with smart technology, could AI achieve the goal of 'de-coaching'?

2. AI Personal Trainers at Home: These projects take 'de-coaching' further by eliminating the need for gyms altogether, offering personalized, professional training at home.

For instance, health startup WorkSmart Labs created Noom, an AI fitness app that uses machine learning to recommend personalized articles, diets, and workouts.

Domestic fitness app Keep also ventured into this space, introducing TOF (Time-of-Flight) camera technology in 2018 to score users' movements in real time, providing professional guidance at home.

Even home hardware manufacturers joined the fray. Huawei's smart screens replicated this approach, using skeletal joint recognition to compare users' movements with professional routines, displaying real-time feedback on TVs.

AI personal trainers aimed to capture the growing home fitness market by offering standardized, real-time feedback to reduce reliance on human trainers and gyms.

However, relying solely on smartphone apps and home screens couldn't fully replace the need for professional guidance, especially for those seeking to avoid expensive training fees.

On one hand, depth sensors and algorithms struggle to provide real-time feedback on movements and intensity, undermining workout effectiveness. Moreover, human trainers play a crucial role in preventing injuries—something algorithms can't replicate.

3. Smart Fitness Equipment with Data Interaction: This led to the rise of data-driven smart fitness gear.

Some are third-party monitoring devices like smartwatches and bands that track workout data and adjust plans accordingly. The Apple Watch, for example, is packed with sensors to collect data on walking, heart rate, sleep, and more.

Others involve smart upgrades to traditional equipment, like resistance bands with touch and tension sensors to track movement accuracy. Crowdfunding projects like Enflux's smart fitness wearables even use full-body sensors for 3D motion tracking, alerting users via an app when their form is off.

By empowering fitness hardware with AI, the goal was to enable data-driven, unsupervised scientific workouts.

These three scenarios nearly exhaust the possibilities of 'AI + fitness.' If smart gyms are limited by lockdowns, why haven't AI fitness platforms and smart equipment gained traction among homebound users? Why do people still prefer following livestreamers for square dancing?

After this episode, human trainers likely won't fear losing their jobs to AI anytime soon. Compared to the hype around smart reminders and AI coaches, real-world fitness preferences remain dominant.

First, the home fitness boom was driven by families, with novices and casual exercisers forming the majority. For them, fun, variety, and simplicity are key.

Terms like 'plank' or 'sit-up' are already daunting, let alone the steep learning curve of smart equipment, which often leads beginners to quit.

Why struggle to figure out which gadget suits you when you can just dance along with a livestreamer?

Second, this was the first large-scale 'professional coaching for the masses' event. Gyms like Super Monkey, LeFit, and Will's launched free online classes to retain members.

Olympic champions also joined the trend. Volleyball players Zhang Changning and Gong Xiangyu filmed a 'Small Moves, Big Health' series, while boxer Zou Shiming, sailor Xu Lijia, shooter Tao Luna, and skater Yang Yang participated in 'Champions Teach You to Exercise.'

The lesson? If professional coaches are accessible, no one cares about saving money with AI.

Third, fitness relies on self-discipline, and creating the right atmosphere is crucial. Live sessions with real-time interaction provide immediate feedback, making it easier to stay motivated.

Gamified products like Nintendo's Ring Fit Adventure, which blends exercise with game rewards, achieved unprecedented popularity.

This offers another insight into AI's role in fitness: post-pandemic, could AI reinvent itself for the offline revival?

Why will offline fitness rebound? For one, free coaching isn't sustainable.

Most livestreams and videos are free, leaving gyms and platforms struggling to monetize online content. For example, Super Monkey's 14-day online camp costs just 399 yuan (about $56), averaging 100 yuan per person—far below a trainer's usual hourly rate.

Once lockdowns lift, coaches will return to offline sessions.

Moreover, 'cloud fitness' has helped many develop exercise habits or buy equipment. E-commerce data shows surging sales of yoga mats, dumbbells, and resistance bands.

How can their impact be sustained? Beyond some individuals transitioning to offline gyms for continued workouts, the integration of online content with offline services will also become a long-term option for gyms surviving through private traffic.

After this round of reshuffling, brands with tight cash flows and weak innovation in services will fall by the wayside, leaving behind a group of user-experience-focused, results-driven enterprises. Will they change the awkward position of AI in the fitness industry?

We don’t yet know. But one thing is certain: the future integration of AI and fitness will inevitably undergo a path transformation.

Here’s a bold prediction:

(1) Abandon the 'Traditional + Digital' Macro Transformation Approach.

Ditch the simplistic idea of adding sensors to rooms, phones, TVs, and equipment. Instead, start from practical experience scenarios to address potential limitations in online interactions.

For example, during online classes, when floor exercises are involved, looking at a phone turns into a 'neck exercise.' Could algorithms automatically recognize and adjust display angles to provide a better teaching experience? This is likely an area where 'AI trainers' need to deepen and refine their capabilities.

Another example: using AI thermal imaging for temperature checks could help gyms reopen more efficiently and safely post-pandemic, reassuring users to return.

Empowering fitness from more practical and detailed angles will be a necessary trend for AI.

(2) Use AI to Manage Human Nature, Not Replace It.

This 'AI Waterloo' has proven that the fantasy of replacing human coaches with AI is impossible in the short term. Factors include user inertia and the necessary constraints of scientific exercise.

VENTO, touted as China’s first AI-powered gym, has closed its Beijing and Shanghai locations. The reason? Pure smart devices couldn’t replace human trainers in providing personalized guidance. While saving labor costs, they failed to attract enough members to sustain operations.

Thus, combining human effort with AI to manage human nature and achieve better workout results might lead to superior training outcomes.

For instance, through sensor data sharing, AI can analyze and identify improper form during home workouts, providing real-time reminders to prevent injuries. Coaches can also use accumulated data to correct, motivate, and design more effective exercise plans—a potentially cost-effective solution.

(3) On the Eve of an Intelligence Explosion, Seek the Digital Harmony.

Currently, the core reason AI fitness software and hardware haven’t taken off is the lack of precision in algorithm recognition, failing to match the expertise of human trainers.

Optimizing the entire system depends not only on developers’ technical capabilities in 3D motion capture and deep learning modeling but also on training and support from large datasets. In contrast, domestic fitness companies—traditional brick-and-mortar gyms aside—have been slow to adopt AI. Keep, an internet fitness leader, boasts over 100 million user fitness data points, while smart gyms like Super Monkey prioritize hardware data collection and integration.

Apple Watch’s health features evolved over years of refinement and patent upgrades before becoming practical. The industry must now refine core algorithms based on data to develop optimal software and hardware with real-world applicability. Only then will AI truly begin to unleash its potential in fitness.

Don’t laugh at AI’s initial failures; its return to the battlefield is inevitable. When spring arrives, we’ll all set out anew.