Journalist Uses ChatGPT to Train for Paris Marathon, Shedding 20 Pounds and Improving Race Times

A Bloomberg journalist has documented his six-month experiment using ChatGPT to develop a personalized training plan for the Paris Marathon, resulting in a 20-pound weight loss and improved race times. The experiment, detailed in a first-person account, offers insights into both the potential and limitations of AI as a fitness coach.

Derek Wallbank, who had previously run the Paris Marathon in 2019 with a time of 4:23:40, set out to improve his performance while preparing for the 2026 race. His approach was straightforward: he asked ChatGPT to create a training plan that would help him lose weight, get faster, and prepare for the marathon distance.

The AI Training Approach

The AI-generated plan focused on several key elements:

• Gradual mileage increase: Starting from a base of 15-20 miles per week, the plan progressively built up to peak weeks of 40-50 miles
• Structured workouts: Including interval training, tempo runs, and long slow distance runs
• Cross-training: Incorporating strength training and yoga to prevent injury
• Nutrition guidance: Basic dietary recommendations focused on balanced meals and proper fueling
• Recovery protocols: Emphasizing rest days and active recovery

Wallbank followed the plan religiously, tracking his progress through a combination of running apps and manual logging. The AI coach provided adjustments based on his feedback, weather conditions, and how he was feeling on any given day.

Results and Outcomes

The experiment yielded measurable results:

• Weight loss: 20 pounds over six months
• Improved race time: Wallbank completed a half-marathon during training in 1:52:30, significantly faster than his previous half-marathon time
• Increased endurance: Able to comfortably run 20+ miles in training
• Better recovery: Reported less post-run soreness and faster bounce-back between workouts

The journalist noted that the AI's ability to provide immediate feedback and adjust plans based on his input was particularly valuable. Unlike static training plans, ChatGPT could modify workouts based on factors like fatigue, weather, or unexpected schedule changes.

Limitations and Challenges

However, the experiment also revealed several limitations of AI coaching:

• Lack of physical assessment: The AI couldn't observe Wallbank's running form or identify potential biomechanical issues
• Injury prevention gaps: While the plan included general injury prevention advice, it couldn't provide the nuanced guidance of a human coach who can spot early warning signs
• Motivation challenges: The AI struggled to provide the emotional support and accountability that a human coach or training partner might offer
• Context limitations: ChatGPT couldn't fully understand Wallbank's work schedule, family commitments, or other life factors that impact training

Wallbank also noted that while the AI was good at creating structured plans, it sometimes lacked the creativity and adaptability of experienced human coaches who can draw on years of working with different athletes.

The Future of AI in Fitness Coaching

The experiment raises interesting questions about the future role of AI in personal fitness and athletic training. While Wallbank's results were positive, he concluded that AI coaching works best as a complement to, rather than a replacement for, human expertise.

"AI can handle the data-driven aspects of training—creating structured plans, tracking progress, and making logical adjustments," Wallbank wrote. "But it still struggles with the human elements: understanding when you're mentally burned out, recognizing when life stress is affecting your training, or providing the kind of encouragement that keeps you going on tough days."

Broader Implications

This experiment comes at a time when AI fitness applications are proliferating, from personalized workout apps to AI-powered nutrition planning. The success of Wallbank's marathon training suggests that AI could democratize access to quality training guidance, particularly for people who can't afford personal coaches or who live in areas with limited access to fitness expertise.

However, it also highlights the importance of human oversight, especially for serious athletes or those with specific health concerns. The ideal scenario may be a hybrid approach where AI handles the data-intensive planning while human coaches provide the emotional support, form correction, and nuanced decision-making that AI currently can't replicate.

As AI technology continues to evolve, experiments like Wallbank's will help define the boundaries between what AI can and cannot do in the realm of personal fitness and athletic performance.

[IMAGE:1]

The featured image shows a runner crossing a marathon finish line, representing the culmination of Wallbank's AI-assisted training journey.