Meaning-Making in Machine Learning Predictions: Lived Experiences of Financial Professionals

Abstract

As machine learning (ML) technologies increasingly shape decision-making in finance, there is growing concern about how professionals interpret and engage with algorithmic predictions. While prior research has focused on model accuracy and system performance, little is known about the lived experiences of users who interact with predictive systems in high-stakes financial contexts. The subjective and interpretive dimensions of this engagement remain underexplored, prompting the question: how do financial professionals experience, interpret, and respond to ML-generated predictions in their daily decision-making processes? This study applies an interpretative phenomenological approach to explore the meanings users construct when working with predictive systems. Using in-depth semi-structured interviews with ten finance professionals and an Interpretative Phenomenological Analysis (IPA) framework, the research identified key themes such as navigating ambiguity, balancing trust and control, emotional reactions to algorithmic uncertainty, and adaptive meaning-making strategies. The findings demonstrate that users interpret ML predictions not merely as data points but as ambiguous and emotionally charged cues that require subjective negotiation within organizational and ethical constraints. These insights provide a deeper understanding of how decision-making is shaped by both the technical nature of algorithms and the human need for meaning and accountability. This study highlights the need for more human-centered AI systems and suggests that future research should investigate interpretive experiences across different professional settings to inform ethical and user-sensitive design practices. Nevertheless, the study is limited by its small sample size and focus on a single professional domain, which may restrict the generalizability of findings. In practical terms, the results emphasize the importance of training programs, organizational guidelines, and transparent system design to better support financial professionals in interpreting and responsibly applying ML predictions.