43.3% of dentists use AI. 22.8% is imaging.

Every dental AI pitch you have taken this year probably opened with a radiograph. That is not an accident of your inbox. It reflects where vendors have put their marketing budget, and it has quietly become shorthand for the whole category: "dental AI" means software that reads X-rays, flags caries, scores bone loss, catches what a tired eye might miss at the end of a long day. It is a real product category, and it works. It is also a small and increasingly saturated slice of what dentists actually report doing with AI.

ADA Health Policy Institute's July 2026 survey of practicing dentists found 43.3% now use AI for at least one task, with another 26.4% planning to (ADA Health Policy Institute, "Dentists Use AI to Make Appointments More Efficient, but Draw the Line at Clinical Decision Making," July 2026, based on a Q2 2026 panel of 589 responding dentists). Imaging and diagnostics accounts for 22.8% of dentists, the single largest reported task. That leaves a wide gap between the headline adoption figure and the category everyone assumes is driving it, and for a practice owner weighing whether a third site can run as consistently as the first two, that gap is the more useful number.

The part of the myth that is true

Imaging is not a manufactured category, and dismissing it would be its own mistake. At 22.8% of dentists, it is the single largest AI use case reported in the ADA HPI's July 2026 data, ahead of insurance verification (13.6%), explaining clinical findings to patients (13.2%), social media (10.7%), business analytics (10.1%) and reception or front-desk check-in (10.1%). It is also the one category with formal validation work behind it. ANSI/ADA Standard No. 1110-1:2025, finalized in early 2025, is the first US standard published on AI in dentistry, and it covers exactly one thing: validation dataset guidance for 2D radiograph image-analysis systems (American Dental Association, "Artificial Intelligence in Dentistry" standards page, 2025). If your practice has evaluated any AI tool this year, there is a good chance it was an imaging tool, because imaging is where the industry has concentrated both its sales effort and its compliance paperwork.

So what this means for your practice: the myth is not that imaging AI is unproven or unreal. It has the largest installed base and the most institutional scrutiny of any AI category in dentistry today. The myth is treating it as a stand-in for the entire category, when the data says it is one slice of a much wider adoption pattern.

What the rest of the 43.3% is actually doing

Set imaging aside and 43.3% still tells a story on its own. Insurance verification (13.6%) and explaining clinical findings to patients (13.2%) are the next two most common current uses, followed by social media (10.7%), business analytics (10.1%) and reception check-in (10.1%). None of those five tasks touches a single radiograph, and together they account for a meaningful share of dentists who are not using AI the way the marketing implies.

The more useful split for a practice weighing expansion is not current use but where the demand is actually headed. Charting and note-taking has the highest planned future adoption of any task at 34.8%, with the ADA HPI reporting no meaningfully sized current installed base behind it. Insurance verification shows planned adoption running at roughly 2.4 times its current use, while imaging's ratio of planned-to-current is closer to 1.1, near its own ceiling (ADA HPI, July 2026; ratio calculated by AI Health Practice from the survey's own current-use and planned-use figures; see our earlier data-drop, "43% of dentists use AI, but not for notes," August 2026, for the full breakdown). Imaging is not where the growth is. It is close to done growing.

So what this means for your practice: if you are benchmarking against "what dentists are doing with AI" in 2026, the honest comparison set is charting, insurance verification and patient communication, not imaging. Those are the categories still filling up, and they are the ones a vendor demo is least likely to lead with.

Why imaging AI will not fix a third site's real problem

A two-site group's actual pain rarely sits in the operatory. It sits in the gap between how site one runs and how site two runs: whether both front desks verify insurance the same way before a patient is even seated, whether both hygiene schedules recall patients on the same cadence, whether a treatment plan explained at one location gets documented the same way at the other. None of that is solved by a sharper caries-detection algorithm. If it gets solved at all, it is solved by the admin-layer categories still sitting under 15% current adoption: insurance verification and patient communication specifically.

That mismatch is worth naming against where the ADA's own governance attention currently sits. The ADA Standards Program's most developed work, the 1110-1:2025 validation standard, applies to image-analysis systems specifically. No equivalent standard exists yet for AI that verifies insurance eligibility or drafts a patient-facing explanation of a treatment plan (American Dental Association, "Artificial Intelligence in Dentistry" standards page, 2025). That is not a reason to avoid those tools. It is a reason to hold every vendor, in every category, to the same standard this site applies throughout: ask exactly what the AI was trained on and validated against, and treat any claim that a tool is "clinically validated" with real skepticism, whether or not a formal standard exists yet for that category. Right now, outside imaging, none does.

So what this means for your practice: before your third-site plan gets pitched another imaging upgrade, ask what share of your own cross-site variance, insurance denial rates, no-show rates, response time to online inquiries, would actually move with an imaging tool. For most two-site groups weighing a third, the honest answer is close to none.

The call

Imaging AI is real, it is the most validated category in dentistry, and it belongs on your list if your diagnostic workflow benefits from it. It is not, on its own, what makes a third site run the way the first two do. The categories with the lowest current adoption, insurance verification and charting, carry the highest unmet demand and the most direct line to the admin consistency problem that actually slows an expanding practice down. Evaluate imaging AI on its clinical merits. Evaluate admin AI on whether it closes the gap between how your sites run day to day. Do not let one vendor conversation stand in for the other.

For an independent, scored read on which category would move the needle first at your own practice, the AI Opportunity and Growth Assessment starts with exactly that question. Book a free 20-minute discovery call to find out where your practice actually stands.

See also: how the GDC's binding consent standard compares to BDA's AI guidance, why multi-site AI standardization is not the barrier it sounds like, and the hallucination rate no AI scribe vendor agrees on.

Related: 43% of dentists use AI, but not for notes · Voicemail isn't a safety net for dental practices · GDC's consent standard is binding, BDA's AI guidance isn't