Exploring AI/LLMs, language and empathy in healthcare organizations from a distributed and phenomenological perspective

Artificial Intelligence (AI) and Large Language Models (LLMs), based on computational AI
algorithms, are discursively promoted to carry the potential of accommodating the problems and challenges that the healthcare sector currently face, such as, information inflation in electronic health records (EHRs), increased workload from digital patient demands and burnout crises among healthcare professionals. Most recently, AI chatbots are claimed to pave the way for better, more empathic communication in healthcare, saving time and resources for healthcare professionals and increasing patient satisfaction. A computational approach to communication and human interaction in healthcare risks, however, to overlook and install a reductionistic understanding of language and interaction processes, as it promotes a disembodied view of human behavior with limited descriptive power. In this paper, we critically outline and appraise these risks and caveats, relying on our own empirical and theoretical work relating to digital health communication, as we point to the situated,
embodied, ecological, and distributed aspects of language and human interaction within healthcare organizations. By doing so, we aim to illustrate the importance of regarding social, emotional, and cognitive processes tied to language and human behavior beyond a computational view. Finally, we offer some concluding reflections that feed into the direction of future research endeavors within the areas of AI, which contributes to cultivating critical awareness concerning technology-mediated communication and behavior in healthcare organizations