Deeper Learning, Human–Machine Interactions and Lessons Learned from the Early Adoption of Artificial Intelligence in Gastroenterology

Artificial intelligence (AI) is rapidly emerging in gastroenterology, with early applications in colonoscopy polyp detection and characterization. Some lessons can be distilled, especially when systems shown to be helpful in randomized, controlled trials did not show similar benefits in pragmatic real-world studies. This article provides a review of available literature and examines the dynamics of human–AI collaboration in gastrointestinal (GI) practices. We outline how deep learning-based AI tools have demonstrated improved lesion detection (e.g., increasing adenoma detection rates in colonoscopy). At the same time, we highlight real-world lessons regarding clinician acceptance, trust, and partnering between gastroenterologists and AI leading to more positive outcomes. Drawing on Kate Darling’s, a robotics researcher at MIT, human–robot interaction theory, we discuss how anthropomorphism and the perceived moral agency of AI influence user trust and ethical considerations. Overall, the integration of AI in gastroenterology shows great promise when humans and machines work in tandem. Gastroenterologists’ experience to date reveals that AI is most effective as an augmentative “second pair of eyes” rather than an autonomous replacement and that successful adoption requires careful attention to human factors, training, and context. These early lessons will inform future deployments and ensure that AI innovations are harnessed to improve patient outcomes in an ethically responsible manner.