Andrew Keeling
Digital impressions are now central to contemporary dental workflows, driven by advances in CAD/CAM manufacturing and intraoral scanning technologies. In parallel, rapid progress in artificial intelligence (AI)-particularly deep learning-has led to its widespread integration into digital impression systems. AI already provides tangible clinical benefits, including real-time artefact removal, improved scan stitching, and more efficient three-dimensional data acquisition. However, these same techniques introduce new risks, most notably the silent modification or completion of scan data in regions of low confidence. Such inferred geometry may appear anatomically plausible while no longer representing a faithful record of the patient. This article examines where AI currently enhances digital impression accuracy and efficiency, and where it may compromise clinical validity. By distinguishing between beneficial data filtering and potentially misleading data inference, the aim is to support clinicians in making informed decisions and maintaining appropriate scrutiny of AI-assisted digital impressions.