SAFE
The core of general dentistry is millimeter-scale work inside a moving, sensitive human mouth — drilling and filling caries, extractions, crown prep, root canals, administering local anesthetic — which no current robotic system performs unsupervised. AI already reads bitewing radiographs for caries and bone loss better than the median practitioner and drafts treatment plans, insurance narratives, and clinical notes, so the diagnostic and paperwork layers compress. State dental licensure plus personal malpractice liability means a licensed human owns every treatment decision and every signature.
Tasks largely resist digitisation. Caries detection on bitewings, periapical bone-loss measurement, and treatment-plan drafting are the tasks AI already does well, but crown margin preparation, extracting a fractured root, locating a calcified canal, and adjusting occlusion by feel with articulating paper are the bulk of chair time and are why this sits at 15 rather than 18 — the imaging and documentation half of the day is genuinely compressible.
Hands-on in uncontrolled environments. Every billable procedure happens with instruments inside a live airway — a patient who gags, bleeds, moves, or reacts to lidocaine mid-drill — and the operating field shifts with tongue, saliva and mandible position, which is uncontrolled in a way a surgical robot on a fixed pelvis is not.
Licensed human required and personally liable. State board licensure, DEA registration for prescribing, and a malpractice policy in your own name mean a paresthesia after a lower third molar extraction or a perforated furcation is your negligence claim, not the practice's software vendor's.
The human relationship is the product. Patients tolerate needles and burs because they have sat in your chair for years and recall bases much of general dentistry on returning families, but a meaningful share of production is walk-in emergencies, insurance-network assignment, and DSO-employed rotations where the patient came for the office, not you — hence 16 rather than 19.
Exists to be accountable for ambiguous calls. Deciding whether a cracked, deeply restored molar gets a crown, a root canal, or extraction with implant referral, and whether an anxious cardiac patient on anticoagulants should be treated in-office at all, are calls with no protocol answer, radiographs that support both options, and consequences you carry for a decade.
Has AI actually changed your work?