Many people track their runs, walks, or gym sessions using wearable devices, yet still struggle to understand what the numbers really mean for their health and motivation. Steps, pace, and heart rate are easy to record, but turning that data into insight that supports long term behaviour change often proves harder. New research suggests artificial intelligence could help bridge that gap by guiding people to reflect more deeply on their exercise experiences. The findings were published in the Information Processing & Management.
The study explored whether a large language model could support post exercise reflection through a guided journaling system. Instead of simply logging performance metrics, participants were encouraged to think about what happened during a run, why it mattered, and how it should shape future plans. The aim was to move beyond tracking towards more meaningful thinking about physical activity and well being.
Researchers developed a system called PaceMind that combines data from wearable fitness trackers with an AI-assisted journaling interface. After each run, users were presented with a structured reflection process covering description, analysis, and next steps. In the AI-supported version, the system generated draft reflections and personalised prompts based on the user’s recent and past exercise data, which users could then edit or expand.
To test its impact, the team ran a two week study involving 21 participants with varying levels of running experience. Each person used both the ai supported journaling system and a more traditional template based journal in a different week. This allowed direct comparison of how each approach influenced reflection, motivation, and exercise behaviour.
Participants reported that the ai supported system made it easier to interpret their exercise data and connect it to personal experiences such as fatigue, mood, or daily routines. Many felt it helped them notice patterns and think more carefully about adjusting future workouts. Measures of perceived reflection quality and meaningfulness were consistently higher when ai support was available, and users also showed greater intention to continue using the system.
But the benefits were not without drawbacks. Some participants found the AI-generated content required extra effort to check, correct, or personalise. This meant the system did not feel significantly easier to use than a standard journaling template. For some, reflecting with ai support shifted mental effort rather than reducing it, replacing the challenge of writing from scratch with the task of evaluating and refining automated suggestions.
Importantly, the study did not find clear short term changes in objective exercise outcomes such as total distance or duration. Instead, researchers observed subtle signs of more thoughtful planning and sustained engagement, such as adjusting goals based on reflection rather than abandoning routines altogether. This suggests ai supported journaling may influence how people think about exercise before it measurably changes what they do.
The findings highlight both the promise and the limits of using ai in fitness and well being tools. While large language models appear capable of supporting deeper reflection and sense making, careful design is needed to ensure automation supports rather than overwhelms users. The challenge lies in balancing intelligent guidance with user control, especially in health related contexts where trust and understanding matter.