We read with interest the study by Djinbachian et al1 comparing in a randomized trial artificial intelligence (AI) versus AI-assisted human (AI-H) optical diagnoses of diminutive colorectal polyps. We found it intriguing, because the authors explored a fascinating facet of the integration of AI into clinical practice: the interaction between humans and machines. The authors argued, echoing findings in earlier publications,2–4 that active human involvement could potentially hinder the effectiveness of the decision-making process in polyp characterization.
We will be happy to hear your thoughts