AI Bias

Dentists across the United States are using AI diagnostic tools to recommend unnecessary procedures at significantly higher rates than independent clinical judgment supports, raising urgent questions about how institutional investors should vet AI adoption in healthcare settings where financial incentives create conflicts of interest. The practice exposes a critical governance gap: AI systems designed to improve care are instead being weaponized by practice management to inflate revenue, creating liability and reputational risk for investors backing dental platforms and service providers.

• Pearl AI promises 37 percent more disease detection than standard diagnostic methods, yet independent dentists rejected AI recommendations as clinically unnecessary.
• A patient was recommended four expensive periodontal treatment sessions by an AI-supported dentist, but three independent dentists disagreed with the assessment entirely.
• Dental office staff reported being pressured by management to act on AI findings or face questioning about why recommended procedures were not performed.

• 37% Additional disease detection Pearl AI claims versus standard diagnostic methods
• 24% More care delivery Pearl AI promises beyond conventional patient assessment
• 4 Periodontal treatment sessions recommended to journalist by AI-supported dentist

Investigative reporting by former Wall Street Journal columnist Joanna Stern has exposed a systematic pattern in which artificial intelligence diagnostic tools are being deployed across U.S. dental practices to recommend procedures that lack clinical justification, driven by practice management financial incentives rather than patient need.

Stern, researching AI applications across industries for her book “I Am Not a Robot: My Year Using AI to Do (Almost) Everything,” documented the problem through direct experience and interviews with dental office staff.

The findings reveal a market failure in which vendors promise significant diagnostic improvements, practices purchase the technology to gain competitive advantage, and then organizational pressure converts those systems into revenue drivers rather than clinical aids.

Pearl AI’s Performance Claims Exceed What Independent Dentists Actually Observe

Pearl AI and competing platforms like Overjet analyze intraoral images and radiographs to identify disease and generate treatment recommendations. The technology operates on principles similar to AI applications already embedded in radiology and other medical specialties, where algorithms assist physicians in spotting abnormalities.

However, the adoption model in dentistry has created a structural problem: vendors market dramatic improvements in detection and care delivery, but the systems generate recommendations that diverge sharply from what experienced clinicians independent of the practice consider necessary.

Stern visited a dentist equipped with Pearl AI, which analyzed her oral health and identified significant plaque buildup, recommending four separate sessions of periodontal treatment at a cost of thousands of dollars. She then sought second opinions from three additional dentists who performed their own examinations without consulting the AI output.

All three independent clinicians rejected the recommendation as clinically excessive.

We see the AI is saying that. But we’re looking, and it’s really not that bad. We think, with some better home care, it can be better.

Independent dentists consulted by Stern

Stern never underwent the recommended treatment. The gap between Pearl AI’s recommendation and three independent assessments illustrates how diagnostic confidence inflates when financial incentives align with positive findings.

Dental Practice Managers Pressure Staff to Weaponize AI Recommendations Regardless of Medical Necessity

Stern’s investigation identified the mechanism driving unnecessary procedures: dental office employees reported that supervisors actively encouraged them to recommend procedures flagged by AI systems. The management pressure was explicit and tied to performance metrics.

Staff described facing questioning from leadership when they failed to act on AI findings, creating an implicit expectation that systems like Pearl AI serve as revenue amplifiers rather than clinical decision support.

This dynamic reflects a broader institutional incentive misalignment common to healthcare settings where technology vendors market algorithmic outputs as objective clinical improvements, while practice operators interpret them as operational efficiency gains, gains measured in procedure volume and revenue per patient rather than in diagnostic accuracy or outcomes.

In radiology and other medical specialties, physician oversight and established clinical protocols create friction against such distortion. In dentistry, where practice ownership is often fragmented across smaller independent and small-group operators, and where board certification and peer review are less stringent, the use of AI diagnostic outputs as management directives has become normalized.

The result is a system in which AI serves as a compliance tool enforcing management directives rather than a clinical tool supporting physician judgment.

Institutional Investors Face Mounting Liability and Regulatory Risk

For investors backing dental AI platforms, practice management software, or dental service organizations that operate chains of practices, Stern’s investigation surfaces a material risk that has not been adequately priced into valuations or due diligence.

If AI systems are being deployed to recommend procedures that lack clinical justification, the liability chain runs through multiple parties: the software vendor for marketing inflated diagnostic benefits, the practice for misrepresenting treatment necessity to patients and insurers, and potentially the investors and equity sponsors who benefit from revenue inflated by unnecessary procedures.

Insurance fraud liability is the most immediate exposure. Dental insurers pay claims based on clinical necessity, and if procedures are recommended and billed based on AI flags that do not reflect accepted clinical standards, the insurers have grounds to dispute payments and pursue recovery.

Regulatory agencies including state dental boards and the Federal Trade Commission have begun scrutinizing AI applications in healthcare more closely, particularly around false advertising of diagnostic capabilities.

A vendor promising 37 percent more disease detection may face FTC enforcement if that claim cannot be validated against clinical outcomes rather than simply comparing detection rates in isolation from clinical significance.

Patient litigation risk also looms. Patients who undergo unnecessary procedures based on AI recommendations that independent clinicians would not have made have grounds for tort claims, particularly if they suffer complications or financial harm from procedures they did not actually need.

Class action discovery in such litigation would expose internal communications between practice management and staff about pressure to act on AI findings, creating direct evidence of misuse.

Regulatory Scrutiny and Clinical Evidence Standards Remain Undefined

Unlike FDA-cleared diagnostic devices in radiology and pathology, which undergo formal validation studies before market entry, dental AI systems have largely avoided rigorous pre-market scrutiny. Pearl AI and other platforms market improved detection rates without peer-reviewed clinical trial data demonstrating that higher detection rates translate to better patient outcomes.

The promise of 37 percent more disease detection is a technical metric, not a clinical benefit, yet marketing materials and sales pitches typically present it as equivalent to improved care.

State dental boards and the American Dental Association have not issued clear guidance on permissible use of AI diagnostic systems, creating regulatory ambiguity that practices are exploiting.

In the absence of clear standards, vendors and practices have defined their own validation thresholds, typically based on internal data or comparison against prior diagnostic methods rather than against independent clinical gold standards or long-term patient outcomes.

Regulators are beginning to recognize the problem. The FTC has opened inquiries into AI-assisted diagnostic tools in healthcare, and state dental boards have begun receiving complaints from patients about unnecessary procedures recommended by AI systems.

Investors in dental AI platforms and practice management software should expect regulatory enforcement actions within the next 12 to 18 months targeting false advertising claims and should conduct immediate internal audits of how their systems are being deployed in clinical settings. The critical forward-looking question is whether vendors will voluntarily implement outcome-based validation standards and clinical governance protocols, or whether regulators and litigants will force the issue. Until then, investments in dental AI face unquantified litigation, regulatory, and reputational risk that current valuations do not appear to account for.