Neither Fear Nor Miracle
How to judge an AI by what it does, not by its black box; "preceptor" machine learning versus "immersion" deep learning.
Steering away from both apocalyptic talk and miracle-selling hype, this chapter proposes a practical criterion for evaluating any AI system in healthcare: judge it by its functional capability and verifiable results, not by the mystery of the "black box". The authors compare traditional machine learning to a resident learning from a preceptor, supervised step by step, and deep learning to a process of massive exposure to cases. These are two distinct ways of "learning" that demand distinct forms of trust, validation and clinical vigilance before they reach the bedside.