From Digital Planning to Physical Execution: A Scoping Review of Physical AI in Dentistry

Objectives

Physical AI in Dentistry is defined as artificial intelligence embodied within a physical platform—robotic, haptic, tactile, or actuator-based—capable of sensing, guiding, constraining, or executing dental and oral health procedures. This scoping review aimed to map the current literature on Physical AI in dentistry, characterize major clinical and educational domains, and identify evidence gaps relevant to future research and clinical translation.

Data Sources

A PubMed search (January 2015 – May 2026) using terms related to dental robotics, autonomous and semi-autonomous robotic systems, haptic simulation, tactile robots, force feedback, and robot-assisted surgery was conducted, supplemented by hand-searching of reference lists and key dental journals.

Study Selection

A total of 191 records were identified. After independent dual-reviewer screening, 68 records were excluded (software-only AI, pure dynamic navigation, TORS in non-dental ENT contexts, AI-free mechanical systems, and borderline records conservatively excluded for conceptual coherence). A final set of 123 studies was included.

Results

The literature is dominated by autonomous and semi-autonomous robotic implant placement systems (n ≈ 90; ~73%), with smaller clusters in robotic endodontics (~13), robotic prosthodontics (~9), robot-assisted oral and maxillofacial surgery (~6), haptic simulation and dental education (~8), and tactile robotic oral hygiene (~3). Evidence derives primarily from in vitro studies, phantom experiments, single-center clinical series, and case reports. Only two randomized controlled trials were identified. Geographic concentration in Chinese institutions is notable (~63% of studies).

Conclusions

Physical AI in dentistry is transitioning from proof-of-concept toward early clinical translation. Robotic implantology is the most mature domain. The field remains methodologically heterogeneous, with insufficient long-term clinical, patient-centered, and cost-effectiveness data. Standardized autonomy classification, reporting checklists, multicenter trials, and regulatory frameworks are urgently needed.