AI Predictions Shape User Experiences in Menstrual Tracking Apps
A new study from arXiv (2605.13261v1) investigates how AI predictions in menstrual cycle tracking apps (MCTAs) influence users' lived experiences. Through 14 semi-structured interviews and a group autoethnography, researchers found that users interpret their bodily and mental states in light of AI predictions, even when those predictions are flawed due to incomplete logging. The interface and AI explanations fail to foster critical awareness of this entanglement, while non-normative users (e.g., those with irregular cycles) report feeling isolated. The paper highlights a self-fulfilling prophecy where AI outputs shape reality rather than merely reflecting it.
Key facts
- Study based on 14 semi-structured interviews and group autoethnography
- Users understand lived experiences through AI predictions
- AI predictions can be faulty due to imperfect logging
- Interface and explanations do not support critical engagement
- Non-normative users report isolation
- Published on arXiv with ID 2605.13261v1
- Focus on human-AI entanglement in MCTAs
- Predictive features provide personalized information about bodies and mental states
Entities
Institutions
- arXiv