“…Multiple challenges were emphasized in the development and implementation of clinical AI, including an absence of ethically defensible laws and policies ( 11 , 33 , 49 , 55 , 57 , 59 ), ambiguous medico-legal responsibility for errors made by AI ( 11 , 22 – 24 , 37 , 48 , 57 ), data security and the risk of privacy disclosure ( 35 , 40 , 54 , 69 ), “black box” nature of AI algorithms ( 57 , 70 ), low availability of high-quality datasets for training and validation ( 57 ), and shortage of interdisciplinary talents ( 11 ). Among the respondents in our survey, the lack of interdisciplinary talents was the primary concern, followed by an absence of regulatory standards and a scarcity in high-quality data for AI training ( Figure 4C ).…”