Welcome to this week’s Dental AI Weekly, honest analysis of where dental AI is going, from someone building in it.
Running a dental practice taught me how quickly a helpful tool becomes a problem when a patient cannot tell who is listening, who is reviewing, or who owns the next step. The technology can work exactly as designed and still leave the patient feeling passed around.
That tension runs through all three stories this week. Dental students are rehearsing hard conversations with AI avatars, rural patients are asking healthcare to stay personal, and Google is testing AI in simulated video visits. Each story points back to the same test: does the technology make human responsibility clearer to the patient?
WHAT HAPPENED THIS WEEK
01 – PRACTICE THE CONVERSATION
Colorado dental students are rehearsing patient conversations with AI avatars.
The University of Colorado Anschutz School of Dental Medicine is using AI-powered avatars to help students practice routine intake, mental health screening, treatment discussions, and other sensitive conversations. The avatars do not automatically reveal every relevant detail, so students have to ask follow-up questions and learn how to uncover the full story.
The school says the simulations are repeatable and adjustable. Students can revisit a scenario, change its difficulty, or practice in another language before working with a real patient. This is training infrastructure, not a substitute for patient care.
What this means for your practice: AI may be most useful before it speaks to a patient. A practice could use synthetic role-play scenarios for treatment explanations, post-op calls, financial conversations, and difficult handoffs. Keep patient-identifiable information out of the exercise. Have a clinical or operational leader check the wording, and change the scenario so the team practices follow-up questions instead of memorizing a script.
02 – TRUST BEFORE SCALE
Some rural patients interviewed by KFF questioned the push for more healthcare AI.
KFF Health News reports that states plan to use part of the $50 billion Rural Health Transformation Program for AI tools, from charting and referrals to risk detection and patient chatbots. The article also describes a limited evidence base, especially in rural settings with constraints on internet access, staffing, and technical support.
Six people interviewed in South Dakota raised concerns about cost, wait times, mistakes, privacy, and losing the relationship with their doctor. A Nebraska rural clinic said its AI scribes reduced burnout and helped clinicians focus on patients. That is one clinic’s reported experience, not proof that scribes improve patient outcomes everywhere.
What this means for your practice: Choose a first use that addresses a complaint patients already make, such as waiting for a callback or repeating the same history. Explain when AI is involved, name who reviews its work, and track the complaint or delay you expect it to reduce. The dentist still reviews clinical documentation. Buying the tool is not the result.
03 – VIDEO CONSULTATIONS
Google tested a research AI system in simulated clinical video consultations.
Google says AMIE, a research medical AI system built with Gemini and Project Astra, can interpret visual and auditory cues, guide virtual physical exams, and reason during a video consultation. A randomized study with patient actors and primary care physicians evaluated history-taking, diagnostic accuracy, management, and communication. Google reports that evaluators assessed AMIE favorably across those measures. The actors preferred AMIE’s video format to its text-chat format. That was a comparison between two AI interfaces, not between AI and a human clinician.
AMIE remains a research system. The study used simulated consultations, and Google says more research is needed before real-world deployment.
What this means for your practice: Score the handoff as carefully as the conversation. A patient-facing tool can sound natural and still leave someone stranded when the question becomes clinical, urgent, or unclear. Test who takes over, what context follows the patient, and whether the patient knows what happens next.
“AI may be most useful before it speaks to a patient.”
BY THE NUMBERS
82%
of respondents to a European Patients’ Forum survey said they would prefer to learn about healthcare AI use directly from their doctor or healthcare professional. In the same survey, 93% said disclosure should happen at the beginning of the care journey or treatment process.
What it signals: Put a short AI explanation into the team’s normal introduction. Say what the tool does, who reviews its work, and how the patient can reach a person. The survey supports direct, early disclosure from a healthcare professional rather than relying only on written notice.
Source: European Patients’ Forum, AI Survey Report 2026. The survey gathered 874 responses from patients and patient representatives across Europe; it is not a U.S. dental sample.
READER Q&A
“We want to send post-visit emails, but writing isn’t our team’s strength. We’ve tried using ChatGPT, but the messages always feel generic and impersonal. Is there an AI tool that can help us write more personal follow-ups without creating more work for the team?”
Practice question selected for Issue 026
BH: I love that you’re thinking about adding this patient touchpoint because most dental offices ignore it.
We train dentists to call patients after surgery, but why should thoughtful follow-up stop there? Every patient should hear that we appreciated their visit, listened to them, and are glad they are part of the practice.
The challenge is making that happen consistently. A team member needs to know what happened during the appointment, what the patient was told, what education or post-op guidance applies, and whether another appointment was scheduled. Then they have to turn all of that into a clear, personal email without misspellings or accidentally saying the wrong thing. Doing that well takes time, and time is something most dental teams do not have.
A generic ChatGPT prompt usually does not solve the problem because ChatGPT was not part of the visit. It only knows what someone takes the time to type into the prompt, which is why the result often sounds generic.
OraCore works differently. Scribe Team drafts the follow-up email from the context of the appointment itself. It can open with a personal detail from the conversation, recap what happened, include relevant education or post-op guidance, and remind the patient about any appointments that were scheduled. The email feels personal because it comes from the actual experience, not a blank template.
We also include reputation management. OraCore scores each visit, and when the experience was positive, it can add a customized request for an online review. Instead of sending the same generic request to every patient, the message can revisit something positive from that appointment and ask the patient to share their experience.
The front office still stays in control. They review the email, confirm that the information is correct, and send it. The AI removes the blank page, gathers the visit context, and creates the first draft. The team provides the final human judgment.
It would be very difficult for a front office team to create that level of connected, personal follow-up for every patient manually. With AI, it becomes a quick review instead of another task that gets delayed or missed.
FROM THE ORACORE BLOG THIS WEEK
Patient Guide | 2 min
What Patients Really Need to Know About AI in Their Dental Care
A plain-language guide to explaining AI without hiding the clinician, the review step, or the patient’s choices.
Communication | 4 min
The Human Element: How Scribe Language Enhances Patient Trust, Understanding, and Case Acceptance
Why the words that reach the patient still need clinical judgment, clarity, and a human voice.
OraCore Scribe drafts the clinical note from the visit. The dentist reviews it before it becomes part of the record.
That’s the week. Reply with what you’re seeing in your own practice. I read every one.
Brad Hutchison
CEO, OraCore AI