S8 E03: Smart AI in Dentistry

[00:00:00] Wright: Everybody's talking about AI, but between all of the hype and the headlines, how do we know what's actually safe, what's legal, and what's worth using? You know, I want to start using AI in practice, but honestly, I am totally not sure what I should be worried about or where to even begin. I'm Dr. ArNelle Wright.

[00:00:21] Ioannidou: And I'm Dr. Effie Ioannidou. Where to even begin? Ugh. Oh, God. I mean, I think that we have already began, girl.

[00:00:28] Wright: Yeah. It's already here.

[00:00:30] Ioannidou: So- It's here ... yeah, it's here. And this is why we, we think that it's really important today to dive in these issues and learn more about the risks, about the opportunities, about the challenges, about the rewards of how AI can be used in dental practice.

So we start from clinical applications, we can move to practice management. I think we will touch a little bit on education which I think is exciting. Yes. I'm looking forward to this. It's gonna be exciting for you. Yes, so let's get started.

[00:01:02] Announcer: From the American Dental Association, this is Dental Sound Bites, created for dentists by dentists. Ready? Let's dive right into real talk on dentistry's daily wins and sticky situations.

[00:01:18] Wright: Hello, hello friends. AI tools are already changing how many dentists diagnose conditions, how we communicate with patients, and how we manage our practice. And AI is gonna continue to transform how we provide care even in the years ahead.

[00:01:34] Ioannidou: I think that you're absolutely right, ArNelle, and I think to help us dive deeper into this topic, we have invited a leading voice on AI in dentistry, Dr. Alexander Lee. Alex, welcome to Dental Sound Bites.

[00:01:51] Lee: Hello, folks.

[00:01:52] Ioannidou: We're so excited to have you.

[00:01:54] Lee: Pleasure to be your guest and be included in a pantheon of greats.

[00:01:57] Ioannidou: Yeah. Oh my God.

[00:01:59] Wright: Oh, out the gate. I love it.

[00:02:01] Lee: Come on. Out of the gate, let's do this. I want to give a big shout-out to my friend Mina Ghorbanifarajzadeh whom you've had on the podcast in season three to talk about this very subject.

So I hope to add a few twists to the conversation because I think she did an amazing job.

[00:02:19] Ioannidou: Definitely.

Wright: This is gonna be really, really good. Yeah. Well, why don't we go ahead and get started, Dr. Lee? Share a little bit about yourself with our listeners and tell us what work you do.

[00:02:29] Lee: Please, uh, it's Alex.

Okay. So I'm a general dentist, I'm a professor, and I serve as the Assistant Dean for Dental Informatics at Western University of Health Sciences, College of Dental Medicine in sunny Pomona, California. This means that I take my clinical skills, academic know-how, and whatever technical aptitude I have, throw it all in a blender and create custom solutions for the custom problems that we have in oral health and oral health education.

My typical day sees me moving between, I don't know, like power chains, power apps, evaluating CVs and CSVs, tweaking X-rays and Xcode. And my team and I have been actively working with AI since 2019. And I guess because of my time in the game, so to speak, I've been Johnny-on-the-spot for AI and the oral health education conversation since 2022, a month before generative AI really landed in our lives.

Um, you know, I've been nonstop giving presentations. I've given 62 AI presentations since that time, published a handful of papers about its use, and I've written multiple policies and influenced a few more from the American Dental Education Association and now the ADA Standards Committee.

[00:03:47] Ioannidou: Oh, wow. That's impressive.

[00:03:49] Wright: Yeah. Very.

[00:03:51] Ioannidou: You may look young, but you are senior in the topic.

[00:03:54] Wright: Mm-hmm.

[00:03:54] Ioannidou: How do you like this, right? Oof. Yeah. No, it's a compliment. But listen, for our listeners, uh, you know, AI is a little bit intimidating still.

[00:04:06] Wright: Yeah.

[00:04:06] Ioannidou: I mean, generationally for sure. So for people that are new to AI, can you help us a little bit understand the landscape and the tools that are available for dentistry now and, and, and how practices are using those tools?

[00:04:22] Lee: Yeah. So in the biggest and biggest of terms, like, you know, when we're looking down from the highest mountaintop, so to speak, um, I think we have to give a lot of thanks to Dr. Manny Chopra, who was then the, uh, chair of the ADA's Council on Dental Practice. And, um, in 2023, he categorized AI use in dentistry into six general buckets while really acknowledging that there's plenty more uses: diagnostics, treatment planning, automation, extended reality, data analytics, and administrative support.

I think most dentists are pretty familiar with AI and diagnostics, specifically imaging and medical reconciliation, and also, like, summarization of medical records. Still, I think a lot of people forget that dentistry was on the forefront of medical use of generative AI. In fact, that's with the proposal button in CAD/CAM restorations.

It's a feature that was introduced in, uh, about 2000, um, and after users complained about how difficult it was to actually make 3D restorations in the CAD/CAM workflow. So AI's everywhere, and dentists have been happily using it for a long time, often without knowing it.

[00:05:40] Ioannidou: It's true. Absolutely true.

[00:05:42] Wright: Yeah. Shout out to dentistry right now for being, like, on the forefront of something, you know?

[00:05:48] Lee: Oh, yeah, and even the imaging stuff all the filters and things like that. You think highlighting caries and making sure all the, all the filters are correct, you think, like, you know, someone at a company is highlighting all that for you? I don't think so.

[00:06:01] Ioannidou: That's right.

[00:06:02] Wright: Yeah. This is so good. So, um, well, before we, we go too deep into it, like, that was, like, a good, like, way to bring us all in with a categorized, um, like the functions and uses of it. Can you talk to us about, like, new tools? We know that it's here. It's not going anywhere. It's been here for a while, but, um, with all of your experience, when something like, comes across your desk, a new tool, um, tell us what questions dentists should be asking first before using new technology, and what approach would you recommend, um, one taking to evaluate it?

[00:06:36] Lee: ArNelle, that's such a great question because, um, this, I think, speaks to a lot of the AI conversation right now and about fundamentals. So for any new technology, regardless of what it is, um, I always ask three very simple questions that every practitioner can ask. It's: can I use it? Should I use it? And how do I use it?

These are the same question that I ask for my electronic health record, my AI-powered clinical decision support tool, my CAD/CAM application, and these questions haven't steered me wrong yet. I'm deeply involved in our institution's HIPAA compliance; I do our automated auditing processes, and I'm pretty key in our accreditation site visit.

So to me, if something is legal, ethical, and technologically feasible, it's usable.

[00:07:29] Wright: There you go.

[00:07:30] Ioannidou: Yeah, absolutely true, and this is what leads me to my next question. You said the key word legal. So where does this legal, uh, and compliance side come in, and what do we need to know to make sure that we are covered?

Uh, I think this is a very important question in practice and in academic dentistry.

[00:07:52] Lee: Absolutely, and I think this speaks to a pretty popular topic that you two had with the legal ramifications in oral health and dentistry. So, um, you know, I- as a disclaimer, I'm, you know, not a lawyer and not even a legal consultant or anything like that.

I'm just a general dentist. But, uh, I do answer this in my can I use it question, and I do wanna tell you all a little secret, is that this is low-key my favorite part of the AI in healthcare conversation to discuss.

[00:08:27] Wright: Ooh.

[00:08:27] Lee: Um, and yeah, it's kinda weird because there's a lot of shenanigans in dentistry, and what I mean by shenanigans, I mean, like, you could be a board-certified specialist and you know your field so well, but then you realize that there's another overlapping specialty that you need to be a master in in order to execute your own.

Like, you think pediatrics and ortho. You think endo, perio. You think anesthesia, oral surgery. You think oral path and oral med, perio, pros. There's no end to the dentistry rabbit hole, especially when you, like, stick a human being on the other end of the mouth. So I call shenanigans on a lot of things there, but laws, especially technology laws, that's finite. Most lawmakers passing technology-related laws aren't experts, and the laws are kind of consensus documents.

And the rabbit hole is deep and circuitous, but it's man-made and it terminates somewhere, most definitively at the Supreme Court. So, um, it's actually my favorite topic because you can actually get to an answer. So on the federal level, there are no laws directly governing AI use in healthcare. So for the most part, AI is governed by existing legal infrastructure we have in place, like HIPAA and the Federal Food, Drug, and Cosmetic Act.

There's a few state laws here and there, but on December twenty, uh, I think like December 13th, 2025, something like that, an executive order came down that charged the Justice Department with challenging state AI laws. And because our government's official stance is that AI should be regulated by the industry itself, um, a lot of laws aren't being activated So, uh, like a few lawyers that you've had on this show, I urge all dentists to be familiar with HIPAA requirements and consult with a legal expert in your state because AI has many different uses, and the exact use, including your own mobilization of the technology, vastly affect the laws that you need to comply with.

[00:10:32] Ioannidou: Now, let me just go a little bit further because I, whenever I have a discussion about AI in scientific publishing, uh, I always pose the question to my audience, do you think that the AI, any form, any, any platform, uh, of, uh, generative AI could be considered an author in a paper? So, and that's, I always propose this question and, and many times, you know, the majority of the, uh, the majority of the audience always believes in humans and themselves.

So they, they seem not to be, to be negative on this. The answer is negative. But there are always a few, um, students or a few, um, scientists that will say, "Yeah, sure. Why not?" So, uh, you know, based on the International Committee of Medical, uh, uh, Journal Editors, my response is always that, uh, AI cannot be considered, any form of generative AI cannot be an author because it's not accountable for the paper.

So I pose the question to you now, um, in the legal frame of this and with the assumption that the industry itself regulates, the industry regulates itself, which is kind of an oxymoron, but whatever, I'm letting this slide. If we have a clinical error in dentistry, in medicine, can be significant and pretty lethal many times, right?

So where does the responsibility, the ultimate responsibility fall, where does it fall, uh, to. Uh, and dentists can come and say to-in their defense, you know, "I, I had this fancy schmancy, uh, AI, uh, directed system as the... And I trust it. It's validated, whatever, and, and I follow instructions." So whose fault is it?

I mean, I think I know the answer, but I pose it to you so we can clarify for our audience.

[00:12:27] Lee: Yeah. So unequivocally, the answer is the practitioner.

[00:12:32] Ioannidou: Mm-hmm.

[00:12:32] Lee: The tool itself, um, cl- typically is called a clinical decision support tool, clinical decision support tool, so, uh, or CDS, and it's exactly what you think it is.

It's clinical decision support tool. Um, there's no company out there right now, at least in the healthcare field, that will say, "This replaces a licensed practitioner." So, um, so that's one. So without a doubt, um, like, you know, whether it's a CAD/CAM restoration that goes south because you just accepted the proposal and you didn't look at it, whether it's a, um, an application that you use to look up drug-drug interactions, whether it's an imaging tool that you look up and try to catch caries and periapical radiolucency, and you don't catch it or you do and it's wrong.

Um, regardless of the situation, unequivocally, it's the licensed practitioner. Now, with that being said, though, there are some laws and some cases that are challenging this. As we know, there are autonomous cars now, and, you know, you have an autonomous car, they're driving, and that is replacing a licensed driver.

[00:13:45] Ioannidou: Sure. Sure.

[00:13:46] Lee: And, um, people are starting to look into this, and there's some debate. There's actually a paper that was published from a graduating law student who said that we should replace the, uh, the legal framework for, uh, for AI in healthcare, because right now, if something goes wrong, it is the practitioner who is, who is at fault.

But if the product is the one that's like, uh, really, really wrong, you can sue the manufacturer, and typically, like that's where, that's, that's where things end. It is arguing for dissemination of liability from not just the manufacturer, but the hospital, the practitioner, the patients themselves even can also have liability.

So it's a really tangled mess right now. But- Short story is the licensed clinician is always on the hot seat.

[00:14:46] Ioannidou: Yeah. I mean, this is, uh, the way you describe it, uh, pretty scary, I have to say. But also perhaps it means that we need more structured frameworks and perhaps a more clear, um, regulatory pathway to follow.

So which right now

[00:15:07] Wright: We don't know how that'll ever happen .

[00:15:08] Ioannidou: Yeah, it's a, maybe it's a voi- I mean, I don't know. It's the Wild West. It really is. Yeah, yeah, correct. It's the Wild, Wild West.

[00:15:14] Lee: Well, but both of you hit on a topic that's actually a very sensitive thing to the American public. Um, for years now since generative AI really made a splash in our lives, if you look at the Pew Research Center numbers, um, it has always been a majority of Americans have, um, have wanted more regulation on AI than less regulation on AI.

[00:15:36] Wright: I can believe that.

[00:15:37] Lee: Yeah. Today it's actually reached a critical point where it's over 80% of individuals like US adults want more regulation on AI, so it's actually pretty significant, and you hit on a really important topic, is that, um, industry regulation is extremely important, but I think the general consensus in the US is that we are looking for federal guidance, for government guidance to help us.

[00:16:04] Ioannidou: Yeah, for sure.

[00:16:05] Wright: We'll be right back.

[00:16:07] Announcer Ad: Your ADA membership opens doors to everything you need to thrive in dentistry, from exclusive savings and career resources to cutting-edge clinical tools and advocacy that shapes the future of the profession. ADA is here for every stage of your journey. Explore wellness programs, continuing education, and solutions that simplify your practice, all designed to help you succeed. Discover your benefits today at ada.org/memberbenefits.

Need CE credits? Get instant access to courses on implantology, sedation, coding, and more. With an ADA CE Online subscription, you will have more than 300 on-demand courses at your fingertips any day, anytime, anywhere. Sign up for your subscription today at ce.ada.org

[00:17:06] Ioannidou: Welcome back to Dental Sound Bites.

Today, we are talking about how AI can be used in the dental practice and the risks we need to be aware of with Dr. Alexander Lee.

[00:17:18] Wright: We, we know that things can be legal, um, and still not right ethically. So while we're talking about it, help us determine where to draw the line on what we should and shouldn't use AI for.

I love how you brought up the clinical decision support tools, like just making sure that we understand that as a category or, or I guess the label. Um, but help us maybe walk through how we decide on that.

[00:17:46] Lee: So this is callback to a previous, uh, answer is, you know, it's can I use it? That's the legal thing, and then now I think we're hitting on the ethical portion of things, which is should I use something?

Just because I can do something doesn't mean I should. Now, in the state of California, I can legally switch lanes on the highway every two seconds as long as it's done safely. But Effie, I know it's like you probably shouldn't do that, especially in Bay Area rush hour traffic, right?

[00:18:17] Ioannidou: I don't think I can do it in the Bay Area rush hour traffic because there is no space to switch lanes. It's packed.

[00:18:23] Lee: This is actually very true too. So societal good, right and wrong, values of not just the profession, but also the individuals. Uh, thankfully, the ADA has its principle of ethics, which says that a dentist has the duty to respect your patient autonomy, act non-maleficently, act beneficently, be just in our actions, and be truthful.

And personally, if my dentist respects me, protects me, puts me in a position to succeed, then goes the extra mile to make sure that what's happening to me is both true and just, I think I would feel pretty comfortable with their use of anything on me, let alone AI, because I believe that these standards apply to composites and grafting materials as strongly as it applies to electronic health records and AI.

[00:19:13] Ioannidou: This is really important, and I really like the, that, you know, the fact that you dop- you brought up the, the policies and the, the principles, um, uh, to, to really, um, stand behind, uh, ethics and morale in, in dentistry, in healthcare in general. So the ADA Health Policy Institute found that about two in five dentists are already using AI.

So, okay, I mean, this is a little bit less than fifty percent. Perhaps from the moment that they, they put this data together to now- Maybe the, the we are, we are reaching five 50 or more than 50%. Because it seems to me that month by month this data is, is changing. But, uh, they may mainly use it, as you mentioned, for imaging and diagnostics, and of course, for, uh, administrative tasks.

[00:20:05] Lee: Mm-hmm.

[00:20:05] Ioannidou: Insurance verification, charting, and all these annoying things, claims, blah. Uh, right? So-

[00:20:13] Wright: Don't even get me started.

[00:20:14] Ioannidou: I, I know. I know that I- it's a soft spot for you. So, so why don't you walk us through the way that dentists are using AI in these areas, and where you are seeing it making the biggest difference, uh, the biggest impact.

And also, uh, do we have any red flags? Where, where are, uh, dentists most likely to overestimate what AI can offer and add to their practice? I know it's a lot, so let's start with, uh- Yeah.

[00:20:42] Lee: Yeah,

[00:20:43] Ioannidou: yeah ... you know, what's the biggest impact that you find?

[00:20:46] Lee: Okay. Well, um, as you said, dentists are already using this however they're- they want to.

And I have found that, um, you know, obviously there's those six categories that Dr. Chopra put out there, but this is where the geeky, like, you know, the technical part of things, like how should I use it question comes into play. So everyone can pretty much cosign on the fact that AI is really good at doing repetitive tasks, okay?

So it is great for repetitive things. So for your imaging, it is great to look at patterns, and it is great to look at patterns in a large bulk sense. Uh, for any given image on any given day, you, me, anybody else, we can outperform an AI algorithm. But over a few thousand repetitions in a very short period of time, uh, we can't outperform AI, even the best of us.

So, um- I think dentists are really focusing in on, uh, using AI to help with repetitive tasks, as you said. There was imaging, there was claims, there was, uh, making appointments, there's answering phones. There's all these things where a few nuances change here and there, but by and large, the actions ultimately are the same.

And a lot of dentists treat these AI applications like their own, like, personal genie lamp. You know, there to carry out their bidding no matter the cost or intention. And this is actually a very critical concern in AI called the black box effect, where you just throw everything in your system and out comes a result that seems correct, but you're not able to fully verify.

So, you know, I guess that kinda wraps up the what dentists are using it for and kinda like You know, what some concerns are because-

[00:22:51] Ioannidou: Yeah, the red flags. Yeah.

[00:22:53] Wright: That's a big one.

[00:22:54] Lee: There's a lot of studies that have found that while most of us are very familiar with AI, the vast majority of us are actually not competent with AI use.

So there's a tool called the Generative AI Literacy Assessment Tool. It's just a simple 20-question survey, and it just focuses on what you know about AI, and the conclusion from a lot of papers that have been written on this is that we're bad at AI. As a population, we don't know when to use it, we don't know how to use it, and we don't even have the proper vocabulary to engage it with.

So it's like a patient trying to debate us on dentistry. You know, can the patient be right? Absolutely. But their lack of training experience, uh, it really puts them at a severe disadvantage. So as someone who's been involved in informatics for over 15 years now, I can tell you that dentists lack the fundamentals to mobilize and fact-check many technologies, including AI.

[00:23:58] Ioannidou: You're absolutely right, and, you know, from, uh, putting the correct prompt and, uh, making sure that you, uh, check on the, on the answers, uh, the, the, on the responses, the, um, the, the phenomenon of hallucination is so frequent- ... in some platforms that a few weeks ago I tested, uh, um, theone platform putting a s- sort of a, a prompt that was associated with my- me.

Um, so I, you know, I wanted to pull some of my, uh, references for a topic that I know that, you know, I've, I've written a lot of papers on end-stage renal disease and periodontal inflammation, and I said, "You know, this is a statement I want to make, and can you please cite, uh, Ioannidou on blah?" Yeah. So, so yeah, rewrites, uh, polishes the statement, whatever, and then adds two papers that I've never written.

Wright: Oh, no. Never.

Ioannidou: It's was complete hallucination

[00:24:58] Lee: And what people don't know about, especially I'm a journal reviewer, and what people don't know about, um, is that, um, a lot of companies, a lot of these publishing companies, uh, journal companies, they don't really like AI company going in and scraping all their information.

Mm. So there's actually, uh, tools in place to help defeat the scraping process. And the, the most insidious part is that it, it doesn't, some of these defense mechanisms are so complex that they allow, like say, 60% of the time to get it right and 40% of the time to get it wrong. So I like to cite a paper from Dr. Alan Turing, his, uh, you know, the, the father of artificial intelligence. And he published his seminal paper in 1950. Yeah, yeah. His computing, uh, you know, forgot the exact name now, but he published his seminal paper in 1950. And, um, I just remember organizing my references and putting, um, you know, putting all that stuff into a generative AI because I wanted to illustrate to the, the people that I was lecturing to what you can do with it at the time.

This was right after, you know, November 2022. And what happened was they, uh, the generative AI at the time had, uh, put in the wrong author. Alan Turing wrote the paper. Everyone knows Alan Turing wrote the paper, but the title was correct, the journal was correct, the link was correct, the time was correct, the author was wrong.

[00:26:37] Wright: Mm.

[00:26:38] Lee: And so people don't know about this, and I've certainly rejected my fair share of papers where the reference ... I have to now check every single reference for every paper that I review because the authors don't line up, the links don't line up, the pages don't line up. So yeah.

[00:26:56] Ioannidou: But it's really important when you give hallucinated, uh, responses like this to correct the, to train the model, right?

Uh, so in my, in my- in this particular example that I gave, I went back and I said, "You know, that, that ... These are not the correct references. Uh, you know, Effie Ioannidou did not write this paper, so send me ..." Uh, and then in the second attempt, the, the papers that the, the platforms, uh, uh, proposed were correct.

[00:27:19] Wright: Oh, good So you are able to train it then?

[00:27:23] Ioannidou: Yeah. I mean- Yeah ... yeah, I train people. So, so but it's in- it's really interesting. And, and, uh, um, how do you think, uh, patients react to the AI becoming part of a dental experience? And, the, do you think the patients lose trust to the dentist? Do you think that the patients gain trust to the system? You know, it's completely different models. So how, how do you see this in practice?

[00:27:52] Wright: This is one of my fears, too. I'm interested in your answer, too, 'cause I, I struggle with this one.

[00:27:56] Lee: Hard-hitting questions, guys. This is- Mm. It's a hard question. No one- Yeah. Yeah. Yeah, this is very, very hard question, and, and I want to hedge my bets on this one and say it depends, 'cause I personally have seen it go both ways.

[00:28:10] Wright: Ooh, can you tell us about both?

[00:28:11] Lee: Yeah, yeah. So as a consumer of healthcare myself, um, I kind of want a second opinion, looking over my doctor's shoulder and making sure they don't forget anything. You know, like, in fact, if I get diagnosed with anything, I don't care if it's caries or, or, you know, some other very bad thing, I want a devil's advocate to be like, "Doc, are you sure? Like, really, really sure?"

And if the doc is able to justify their stance and go, "Yeah, you know, I am really, really sure," that gives me a lot of comfort as a patient. Uh, you know, 'cause, and I think that the other way when things go south is when a clinician doesn't know how their technology works and is not up to date on evidence-based techniques.

Like, for example, I've seen a lot of dentists have 3D printers sitting in their waiting room. Um, because they think... No, no, uh, they think that it makes them look high tech.

[00:29:07] Wright: It's attractive.

[00:29:08] Lee: It's attractive, makes them look high tech. Mm-hmm. No, this is actually- Mm-hmm ... a very bad idea 'cause the 3D printers throw off cancer-causing particulates, okay?

And you're not supposed to have that there. So FYI, if anybody listening in, you have your, you have your 3D printer in your waiting room, get it out of there now. So like, just if you don't know how your AI is checking drug-drug interactions, why should a patient have confidence in your recommendations?

'Cause you're just regurgitating what they can find themselves. And as a society, we've been trained to always have the answer. Like educationally, we get rewarded for answers, and the faster you come up with the answer, the bigger the reward. That's how school works. That's how game shows work. Even in our normal discourse right now, people wanna say, "Me, me, me, me, me."

And so for healthcare, I think the clinician needs to be confident enough to sometimes hit the brakes and say, "Hey look, we are in control of the situation. We don't have answers yet, but our next steps are XYZ so that we're sure we have the answers." So I've just seen patients get concerned because the clinician seems very out of control about the situation, and it's the AI or it's the system that's steering them, not the clinician.

[00:30:25] Wright: From your experience or some of the stories that you've heard, have those clinicians, um, have they shared with the patients that, "Oh, this is like from AI"? Like how, how does, how does the patient become confused about, you know, where those answers came from? Is it just because they know that the office is, is utilizing AI for the majority of their, I guess, answers?

[00:30:48] Lee: Well, I think you have to look on social media for a lot of these questions. Um, so there is a dental side of a very, very popular Gen Z social media platform.

[00:31:00] Wright: Okay.

[00:31:01] Lee: And there are many examples of AI, generative AI content like post-op instructions, uh, gone wrong. So, um, there are now a lot of these AI applications where you buy, they help triage your phone, uh, help triage your patients for after-hours emergencies because for a private practitioner, it's sometimes can be very hard to get someone the information they need and especially after hours.

You know, it's like we all have lives, hopefully. So yeah, there are times where things go very sideways and, yeah, that can really erode a patient's trust.

[00:31:44] Wright: Mm-hmm.

[00:31:44] Lee: Uh, especially when you're not upfront or especially when it's like, hey, you know, like you don't tell the patient, "I will try to get back to you, and this is a measure of last resort.

Like, if you need something right away and no one's able to answer, this is something you can use. But I really recommend that, you know, you talk to somebody or you speak to somebody." You know, it's-- I think it's just setting those expectations down because I've-- Because, you know, just think about it as a patient ourselves.

[00:32:13] Wright: Yeah.

[00:32:14] Lee: You know, I, I don't like the rug being pulled from under me.

[00:32:17] Wright: Yeah. Well, so as we talk about, you know, I guess things going sideways, um, talk to us about some blind spots that dentists should be checking before bringing AI into their offices, because it takes time to develop an expertise in using AI.

Um, it just doesn't happen overnight. So what are some blind spots? Um, we mentioned a few red flags of, you know, different scenarios, but if you can just walk us through some blind spots that, um, dentists should be checking, that would be really, really useful.

[00:32:49] Lee: Yeah. I, I hate to keep harping on it, but legality is the biggest red flag, 'cause I don't care how great a piece of technology is, if it's illegal to use, don't use it.

If a vendor doesn't have a business associates agreement or BAA, if your office doesn't have a security risk assessment or SRA, if you have no documented protocols on how and when to use something and how to train someone to use it, if you can't or don't know how to audit your technology regularly, and if you don't know if something is FDA-approved or cleared for its purpose, then you probably shouldn't be using it.

Every conference I go to, I end up being roped into a meeting with a new vendor or enthusiastic innovator- and they're just showing me the latest and greatest toys, and most of the time, these individuals do not have their legal bases covered. They're so focused on the cool features, they don't have the basics down.

I've had vendors tell me that you don't need a BAA when working with protected health information. No, that is false and a federal crime. I have had vendors not know the difference between approved, FDA approved and FDA cleared. I've even had vendors come up and tell me, "Hey, this software is in beta, so you don't need to sign anything. Use at your own risk."

And by the way, in US law, if it acts like a clinical tool, works like a clinical tool, and is used as a clinical tool, it's a clinical tool, no matter what stage of development it's in, and legal agreements need to be in place. So I think dentists aren't confident enough to ask the vendors for the legal support and legal infrastructure that is necessary.

[00:34:33] Ioannidou: You know why they are not confident enough? And this takes me to the next question Because they are not trained. And here is the dental education piece, right? How behind we are in our dental school education, and, you know, we like it, we don't like it. So there is a gap here, right? So we, we... the training of the, I mean, and I'm not implying that there are definitely dental schools that have incorporated AI in their curriculum, but the majority of them, no, they haven't.

So, um, the way that the AI is changing the way the, how the students and early career, uh, dentists learn, what do you think? How should we adapt the dental educational, uh, models? Uh, and now, I know it's very difficult question what I'm asking, but think that the- think, think about the demographics of people in dental school, think about the demographics of full professors, think about who is teaching in these clinics.

Frequently, you find people that have retired from private practice and moved to, uh, dental education. Frequently, you find part-time faculty that, uh, like volunteer their work in clinical practice. So taking, uh, all this under consideration, you solve the puzzle, Alex.

[00:35:46] Lee: Wow. Wow. Okay. Uh, it's like, all right, back to you guys.

[00:35:51] Ioannidou: Yeah.

[00:35:52] Lee: Uh, no, this, as, as an educator for 16 going on 17 years now, um, in oral health, this is the hugest concern of mine because as licensed clinicians, we have a small reprieve from AI, and it's because we have a piece of paper that says that, well, um, if anyone or anything else tries to do this exact thing, they're gonna get thrown in jail.

You know, our dental license. Um, but everything else in society Uh, there's, there's none of that protection.

[00:36:25] Ioannidou: Sure.

[00:36:25] Lee: And sometimes, like, that's good and sometimes that's bad. So like, for example, for me, I'm terrible at home improvement stuff. Hate it, have no interest in it, am totally incompetent with it, and I'm happy that AI has risen my home improvement game, but I'm not interested in achieving any more, okay?

This doesn't work, though, when the name of the game is improvement, like education. This doesn't work when you're a healthcare provider and someone's life in your hands. You need to be curious. You need to strive for more, and you need to test yourself and your thoughts. And I don't know if like, you know, like if you don't know how to write something and you rely on AI to write for you, you'll only ever sound as good as an AI.

And so if we're being really honest, that's neither great nor really engaging nor very genuine. Like, uh, I'm sure you all get emails right now that sound like pathologic people pleasers, you know?

[00:37:31] Ioannidou: Exactly. Yes. Yes.

[00:37:33] Wright: Yes.

[00:37:34] Lee: So I, I think that it's increasingly apparent that if you didn't arrive at the generative AI scene with a skill, you're going to be using AI as a crutch, and that AI will be your skill cap.

So, um, you know, I just emphasize to my students that you need to learn the craft and the fundamentals, and when you learn the fundamentals, you can extrapolate from there.

[00:37:59] Wright: Yeah.

[00:37:59] Lee: Like, I can't have a conversation with someone about CBCTs and digital workflows if they don't know what DICOMs are. And yes, I have had this conversation.

I'm like, "Do you know what a DICOM is?" "No, but that doesn't matter in this conversation." I'm like, "No, this is exactly why it matters." And so you don't start at the end where you know how to do the task and you're just looking to make it more efficient. You start with learning how to do the task first and be curious about it. And my fear is that AI will be the death of curiosity-

[00:38:31] Ioannidou: Oh, yeah

[00:38:31] Lee: because everything will be good enough when it clearly isn't.

[00:38:35] Wright: True.

[00:38:35] Lee: Right? I mean, for what it's worth, in a recently unsealed court document from Microsoft's internal documents on a lawsuit, uh, the, an exec, uh, states that generative AI has created a doom loop for the internet, where because AI has ingested everything on a, on a site, and people actually stop clicking on the site and accept that the AI output is good enough, and there's numbers from Microsoft that back this up, and this is a matter of public record.

[00:39:06] Wright: Wow. Oh. Your description just reminded me of, like, I guess, I don't know if this is accurate or not, but I'm still gonna share it, like, the transition from, like, analog to digital. It's kinda like a lot of people just wanna jump to digital everything. Although it's, like, you know, faster, more efficient, you know, there's accuracy, all these things, um, you kinda need to know how to do the things in analog too before you jump, and understand, you know, analog before you jump to digital.

Would you say?

[00:39:33] Lee: Yeah, completely. Like, for example, for today in our podcast, most people when they do any, uh, audio setup for a podcast, they will just get a USB mic and plug it in. Um, and if it goes wrong, they don't really know how to fix it besides keep updating and keep doing stuff. Well, I'm actually using a analog setup right now.

I don't know if I can-- you can see it, but I'm actually using an analog setup, and it is an analog signal to a digital interface. And I know how all this stuff works, so I can, based on my room, based on my demands, based on the noise rejection you need, because I have a creaky arm on my chair, I'm able to select the right microphone and connect it to my computer.

[00:40:18] Wright: There you go.

[00:40:19] Lee: So yeah.

[00:40:19] Wright: You're fancy. So fancy. You're fancy, Alex. Well- We, we have a team of people who help us with these things, so. Um, well, I wanted to ask one other thing, which was about just kind of giving us a glimpse of the future if you could. Um, what AI capabilities do you think are gonna become super common in dental practices, and maybe some trends that you think we should start paying attention to today?

Maybe just look, like, five years down the road.

[00:40:47] Ioannidou: Can I just interject here and say that there was, uh, just a paper on... Was it on The Guardian? Where a security executive in one of the big companies, AI companies just, uh, uh, revealed to the rest of us that by 2030, you know, we will be pretty rare as human beings.

So that's my, that's my crystal ball, but I don't want to be, uh, pessimistic. So you go, Alex.

[00:41:16] Lee: You're not wrong in, in greater society, but we're talking about oral health- Dentistry ... like I have a l- I have a lot of, uh, fears about AI and the ethics of it and the legality of it and how it's gonna be used.

But specifically for healthcare and for dentistry, oral health, I think all you need to do is you just need to look at the next lane over and see how medicine's doing.

[00:41:41] Wright: Mm.

[00:41:41] Lee: Like in 2025, dental AI venture capitalist spend was an estimated $500 million. Can either of you guess how much the overall healthcare spend for venture capitalists in AI for healthcare was?

[00:41:57] Wright: It has to be way, way more than that.

[00:41:59] Ioannidou: I mean, probably billions.

[00:42:01] Wright: Yeah.

[00:42:02] Lee: Okay. How many billions? Eight? Three

[00:42:06] Wright: 10

[00:42:07] Lee: 10? Okay.

[00:42:09] Wright: Probably even more. I would say

[00:42:10] Ioannidou: five.

[00:42:11] Lee: Five. $18.2 billion. Oh, wow. Ooh.

[00:42:16] Wright: Yeah.

[00:42:16] Lee: So dental venture capitalist AI spend is 2.7% of the overall AI healthcare spend. So- we are a drop in the proverbial ocean.

And so medicine has always had the latest and greatest AI toys. Sure. And they've been at it for much longer at a much higher level than us.

[00:42:40] Ioannidou: Are you surprised? You're not surprised. That's history, right? Yeah.

[00:42:45] Lee: Yeah. But you look at any of their electronic health records, um, how group practices are structured.

Even look at the curriculum. You even look at the medical curriculum, and Effie, you might, you might, uh, appreciate this. If you look at how, uh, medical medicine. Uh, have you seen how accreditation works for, uh, medical programs?

[00:43:05] Ioannidou: Not to the depth of CODA because we are going through accreditation next year, early next year.

So I'm like- Okay ... in the CODA mode all the time, 24/7. So don't talk to me about accreditation. I have an allergic reaction.

[00:43:17] Lee: Hey, I'm, I'm doing the same thing as you are too. And let me tell you, for a lot of the metrics, there is literally a button in, in accreditation, uh, software for medicine where you hit a button and it uploads everything into a portal for medicine.

[00:43:32] Ioannidou: I love this. Can I get it?

[00:43:35] Lee: Yeah. Yeah. So that's how far ahead they are- in all measures of, of that yardstick. So yeah, if you, if you want a if you want my crystal ball, and I'm willing to give it to you-

[00:43:47] Wright: Mm-hmm ...

[00:43:48] Lee: just look at where medicine is.

[00:43:50] Wright: So good.

[00:43:52] Announcer: On the next Dental Sound Bites.

[00:43:54] Wright: I'm so excited, Effie. Today we're joined by the new president of the American Dental Association, Dr. Tom Paumier.

[00:44:01] Paumier: We can do anything we want, but we can't do everything we want.

[00:44:07] Wright: Pay attention, people.

[00:44:09] Paumier: There's a lot of challenges that are in front of us as a profession and as the ADA, and those headwinds and the things that I hear the most, not just from travel, but on message boards and, uh, from all dentists, members and non-members-

[00:44:21] Ioannidou: And scary. Some of them are scary. Yes, a lot.

[00:44:24] Paumier: But the most, uh, challenging leadership issues that I've always seen at every level are not related to the issue itself.

[00:44:32] Ioannidou: I love this.

[00:44:33] Paumier: Every generation says, "Oh my gosh, I wish I had it the way the past generation was." That was the golden age of dentistry. The golden age of dentistry is always in front of us.

I believe that.

[00:44:46] Wright: Okay, one last question before we let you go. Um, is there anything that we didn't cover, anything else you wanna share with us that we didn't ask? For example, like, what's one thing that every dentist who's listening to this episode can do, uh, tomorrow morning to become more informed about AI and dentistry?

[00:45:05] Lee: So main takeaway is always just go back to the fundamentals of technology implementation. There's no federal AI healthcare laws right now, so if you wanna use AI in your practice, the same laws apply to your electronic health record, to your imaging application, to your CAD/CAM system. And if you don't feel like you've done a good enough job where you have no issues with a HIPAA auditor coming in tomorrow into your practice, then get to work.

And there's actually tons of resources for you out there, from the ADA's own standards documents to, um, even the government has a cybersecurity, uh, site. It's called the 405(d) Program, that gives you training materials, standardized training materials for you and your staff, and to incorporate into your own, uh, into your own practice manual.

And also, um, the government also gives you a security risk assessment tool. I personally know a lot of cybersecurity healthcare experts who- whose entire business plan is to use this free government tool to give you recommendations for your practice.

[00:46:13] Wright: Nice.

[00:46:14] Lee: So it's all out there for you.

[00:46:16] Ioannidou: Listen, this was amazing. We thank you so much for making the time to be with us and, and really enlighten us. So this is, uh... This was lovely. Thank you so much, Alex.

[00:46:28] Lee: Well, thank you. Thank you, ArNelle. This has been absolutely wonderful.

[00:46:32] Wright: So good.

[00:46:34] Ioannidou: And a reminder that we'll, we will have all these, uh, resources and information mentioned in the episode at the classic show notes on the ada.org/podcast.

[00:46:49] Wright: Yes, and don't forget, if you know someone who could use the information that's mentioned in this show, don't forget, please share it with them.

[00:46:55] Ioannidou: So you can follow us on all ADA social channels, and hit subscribe wherever you listen so you never miss an episode. You can also rate this episode, preferably high, and write a review. It helps more people find our show.

[00:47:10] Wright: Bye.

[00:47:12] Ioannidou: Goodbye.

[00:47:16] Announcer: Thank you for joining us. Dental Sound Bites is an American Dental Association podcast. You can also find this show, resources, and more on the ADA member app and online at ada.org/podcast.