Dentists already decided what AI is for.

Start with the headline. More than two in five dentists surveyed by the ADA use AI for at least one task 1. Now look at any specific operational job. “Around one in ten dentists use AI for business analytics (10.1%), social media (10.7%), and reception or front desk check-in (10.1%)” 1. Insurance verification does a little better: “More than one in ten use AI … for insurance verification (13.6%)” 1.

That is the gap I keep coming back to. Wide adoption of “at least one task”, thin adoption of any particular task. The ADA does not frame it that way. I do. When four in ten say yes to AI and roughly one in ten say yes to any given business job, my reading is that most practices are using AI somewhere, lightly, rather than running it anywhere, fully.

The plans point to the same place. “Dentists indicate that their future AI use will most likely revolve around tasks generally related to administration and practice efficiency” 1. Specifically, “around one-third plan to use it for charting and note taking (34.8%) and insurance verification (32.6%)” 1. Among dentists who plan to adopt AI, the top planned uses are “front-desk functions such as check-in and appointment scheduling tools” 2. The operations side is where the intent lives. I cover the dullest item on that list in “The most boring AI use case is the one dentists actually want.”

Not everyone is coming along. “Another one-quarter indicate they plan to use AI in the future, while three out of ten indicate they never intend to use AI” 2.

As a cross-check outside dentistry alone, CAQH reports that “More than half of plans and one-quarter of providers now use AI, with provider use focused on administrative tasks such as eligibility checks, note-taking, and patient communications” 5. That index covers medical and dental together, so it is a broader picture, but the pattern is the same: the work is administrative.

This is landing in a tight market. “Two out of five dentists reported in Q3 that they are not busy enough and could treat more patients” 3. The ADA notes that “Dentists continue to face a ‘fiscal squeeze’ within their practices” 3. It adds that “Decreasing dental spending may be reflective of fewer patients going to the dentist, fewer patients accepting treatment plans, and lower reimbursement” 3.

Asked for their top three practice challenges, dentists named staffing, recruitment and retention (62.2%), low insurance reimbursement and denials (57.7%), and rising expenses and overhead (45.7%) 4. “Around one-third (32.4%) of dentists report being overworked” 1. And “Dental practices are struggling with ongoing staffing shortages, and dentists are spending more of their work week on non-clinical tasks” 1.

Put those together. Not busy enough 3. Reimbursement under pressure 3. Overhead among the top challenges 4. More of the week spent on non-clinical work 1. That is the context in which AI is being adopted, and in my view it explains the shape of the adoption.

The ADA observes that “some dentists may view AI tools as a workaround to ongoing staffing problems” 1. I think that is exactly what is happening, and I think it is the wrong frame.

Dentists have already drawn the clinical line themselves. “There is a clear dividing line, so to speak, where dentists seem open to AI” 1. On the far side of that line, “The hesitancy is driven in part by dentists’ fears of AI replacing human judgment” 1. Good. That line is theirs to draw. It also tells you where the work is: on the business side, where no line has been drawn at all.

So, what is the best AI for a dental practice? Not a product. The best AI is the one that runs inside a defined process and can prove what it does. The FTC puts a simple question to companies selling AI: “Are you exaggerating what your AI product can do?” 6 Put it to every vendor, and to your own office.

Can an AI receptionist work for a dental office? It can, if “receptionist” is a mapped job with a handoff, not a voice placed in front of a mess. Will AI replace the dental front desk? I will not make a jobs forecast. What I will say is that the front desk is a set of processes, and the ones that are mapped and measured keep running whoever or whatever executes them.

Name the job. Not “use AI at the front desk”. One job: new-patient calls, recall, insurance verification, reschedules.

Map the process. Who or what does each step, what gets handed off, what counts as done.

Keep humans where judgment is needed. Dentists already decided that for clinical work 1. Do the same on the business side, on purpose.

Measure it. If you cannot say what the job produced, you do not have a system. You have a subscription.

Demand proof from vendors. The FTC’s guidance is blunt: “You need adequate proof for that kind of comparative claim, too, and if such proof is impossible to get, then don’t make the claim” 6. A vendor that cannot show you proof is asking you to carry the risk. On privacy, ask whether the vendor signs a business associate agreement. A “HIPAA compliant” badge is not a government seal: “OCR does not endorse, certify, or recommend specific technology or products” 7.