This study explores how different data representations support reflection in personal informatics. We compare text generated with a Large Language Model (LLM), a standard chart, and the combination of both. Our findings reveal that users experienced more reflection, focused attention, and reward when presented with the generated qualitative data compared to the standard charts alone. Our work demonstrates how automatically generated data descriptions can effectively complement numeric fitness data, fostering a richer, more reflective engagement with personal wellbeing information. We conducted an online survey with n = 273 participants, using step data from fitness trackers and comparing three presentation formats: standard charts, qualitative descriptions generated by an LLM, and a combination of both. Our findings suggest that text-based descriptions fostered deeper reflection through comparison, commanded more focused attention from the users, and made the users feel that the act of viewing personal data was worth the effort. We offer three contributions to the HCI community: (1) an interview pre-study that uncovers how users feel about transforming fitness data into stories and their receptivity to automatically generated descriptions; (2) an online experiment conducted on a dedicated platform to assess how these AI-generated narratives influence reflection on and engagement with the users' personal data; and (3) insights on the potential of generative AI to improve the quality of reflection in PI systems. The study found that text-based descriptions fostered deeper reflection through comparison, commanded more focused attention from the users, and made the users feel that the act of viewing personal data was worth the effort. The results suggest that text-based descriptions are more effective than standard charts in supporting reflection and user engagement with personal data. The study also highlights the importance of designing narratives that are neutral, concise, and grounded in the data, while avoiding overly positive or negative tones that could trigger rumination. The findings suggest that AI-generated narratives can be a valuable tool for enhancing the reflection and engagement with personal data in PI systems.This study explores how different data representations support reflection in personal informatics. We compare text generated with a Large Language Model (LLM), a standard chart, and the combination of both. Our findings reveal that users experienced more reflection, focused attention, and reward when presented with the generated qualitative data compared to the standard charts alone. Our work demonstrates how automatically generated data descriptions can effectively complement numeric fitness data, fostering a richer, more reflective engagement with personal wellbeing information. We conducted an online survey with n = 273 participants, using step data from fitness trackers and comparing three presentation formats: standard charts, qualitative descriptions generated by an LLM, and a combination of both. Our findings suggest that text-based descriptions fostered deeper reflection through comparison, commanded more focused attention from the users, and made the users feel that the act of viewing personal data was worth the effort. We offer three contributions to the HCI community: (1) an interview pre-study that uncovers how users feel about transforming fitness data into stories and their receptivity to automatically generated descriptions; (2) an online experiment conducted on a dedicated platform to assess how these AI-generated narratives influence reflection on and engagement with the users' personal data; and (3) insights on the potential of generative AI to improve the quality of reflection in PI systems. The study found that text-based descriptions fostered deeper reflection through comparison, commanded more focused attention from the users, and made the users feel that the act of viewing personal data was worth the effort. The results suggest that text-based descriptions are more effective than standard charts in supporting reflection and user engagement with personal data. The study also highlights the importance of designing narratives that are neutral, concise, and grounded in the data, while avoiding overly positive or negative tones that could trigger rumination. The findings suggest that AI-generated narratives can be a valuable tool for enhancing the reflection and engagement with personal data in PI systems.