As wearable devices like the Apple Watch become increasingly complex in tracking personal health metrics, a recent experiment reveals a critical caution regarding the use of artificial intelligence in interpreting that data. A *Washington Post* technology columnist found that submitting raw health data to an AI chatbot,specifically ChatGPT,resulted in a false and alarming cardiovascular diagnosis. This incident highlights the potential for misinterpretation when complex datasets are analyzed without proper context and underscores the continued need for professional medical evaluation.
Smartwatches, like the Apple Watch, have become ubiquitous tools for tracking personal health and activity levels. However, a recent experiment highlights the potential dangers of relying on artificial intelligence to interpret that data.
A technology columnist for The Washington Post discovered that feeding raw health data to an AI chatbot, like ChatGPT, can lead to inaccurate and even alarming diagnoses.
The experiment began with exporting health data from an Apple Watch. The data, a complex XML file containing thousands of data points, was then submitted to ChatGPT with a simple prompt: identify any unusual patterns or potential health concerns.
The AI’s response was startling, falsely indicating a serious cardiovascular issue. ChatGPT flagged what it perceived as extreme drops in heart rate, suggesting the user’s heart was stopping or slowing dangerously.
Fortunately, the columnist didn’t immediately panic. A manual review of the XML file revealed the AI’s diagnosis was entirely incorrect. The issue stemmed from ChatGPT’s inability to correctly interpret the technical structure of the Apple Health data.
Specifically, the AI misinterpreted “zero” values and empty spaces within the XML file – which are used as separators between tracking sessions – as indicators of a failing heart. This highlights the challenges of applying AI to complex datasets without a thorough understanding of the underlying data structure and context.
The columnist’s heart was, in fact, perfectly healthy. The incident underscores the importance of professional medical evaluation and the limitations of current AI technology when dealing with sensitive health information.
Baca juga: TWS Apple AirPods Pro 3 Resmi, Bisa Monitor Detak Jantung dan Terjemahan Langsung
This case adds to a growing body of research examining the potential for “hallucinations” and inaccuracies in AI chatbots. OpenAI research has recently explored the underlying causes of these issues, emphasizing the need for caution when relying on AI-generated insights.
