COMUNICAÇÃO DE MÁS NOTÍCIAS E A INTELIGÊNCIA ARTIFICIAL NA MEDICINA: Uma revisão de escopo

ABSTRACT: INTRODUCTION: Communicating bad news is challenging in everyday clinical practice; the incorporation of artificial intelligence (AI) offers opportunities for ethical and effective optimization but raises ethical concerns about preserving empathy and humanization in the physician–patient relationship.METHODS: A scoping review was conducted according to PRISMA-ScR (2018–2024) in the PubMed/MEDLINE, SciELO, and BVS databases, with no language restrictions. Four search strategies retrieved fifteen unique records; eleven studies were selected after dual independent screening of titles, abstracts, and full texts. RESULTS: The narrative analysis identified five main strands: Symptom checkers that enhance autonomy in initial triage but lead to unnecessary referrals. Large language models used in radiology reporting and patient messaging, which speed up report drafting but function as “black boxes,” requiring human review to maintain an empathetic tone. Scientific writing tools that improve manuscript fluency without replacing subject-matter experts. Hybrid protocols (SPIKES AI) that show promise but have been little tested in real-world settings. Ethical challenges such as embedded biases, model “hallucinations,” and lack of transparency. CONCLUSION: AI can alleviate the technical workload of delivering bad news but cannot replace empathetic human interaction. We recommend developing conducting randomized trials to assess impacts on empathy, satisfaction, and patient outcomes in both the short and long term.