Wearables have transformed the way we track our fitness and health. But beneath their sleek designs, these devices contain a catch-all solution to provide users with actionable insights – AI summaries. However, after testing various wearables, I’ve come to realize that these AI-generated summaries often fall short of expectations.
These platforms claim to utilize artificial intelligence to relay raw workout data into plain English. Yet, the typical morning summary tends to be superficial and unhelpful. For instance, Strava’s Athlete Intelligence feature might say you slept well with a resting heart rate in line with your average value – but what does that mean for your overall health? Similarly, Oura Advisor provides some useful insights, like suggesting rest when stressors like heat are present. Nevertheless, it often struggles to delve deeper and provide actionable advice.
The issue lies in the limitations of Large Language Models (LLMs) and the messiness of private health data. Wearables like Strava and Whoop lack the comprehensive health data points necessary to create holistic insights. This results in summaries that feel like repackaged data or book reports written by a fourth-grader relying on Wikipedia.
While some users find AI features helpful, I believe it’s essential to acknowledge these limitations. The current state of wearables’ AI summaries is not worth paying extra for. Instead of relying on duct-taped solutions, we need more substantial efforts to develop intelligent and personalized insights that go beyond surface-level data analysis.
Source: https://www.theverge.com/fitness-trackers/694140/ai-summaries-fitness-apps-strava-oura-whoop-wearables