SPEAKER_00 0:27

What's up everyone? This is Dr. Tjaeyan. A couple of weeks ago, I was at MIT studying AI, healthcare, governance, and data security. One of our sessions was led by a Harvard law professor named Glenn Cohen. He is one of the top experts in the country on health law and medical AI. And I walked out of that room with one thought stuck in my

Why I Avoid AI Charting

SPEAKER_00 0:53

head. I run six companies, almost all of them powered by AI. And I still will not let AI touch my patient charts. Well, not yet. Now that probably sounds strange, coming from the guy who keeps telling you to use more AI. But stick with me. By the end of this video, you are going to know exactly where AI is dangerous in your practice, where it is completely safe, and the one rule I use to tell the difference. So are you ready? Let's get into it. Well, here's what's happening right now. Every single week, somebody is in your inbox or your feed selling you AI for the clinical side of your practice. I know that. AI scribes that's right now very popular. AI scribe that listens to your visit and writes the note for you. Well AI that reads the chart, AI that suggests the diagnosis, and listen, I get the appeal. You are

The Big Push For AI Scribes

SPEAKER_00 1:53

tired, you are behind on your notes, and somebody is promising to take that pain away. But here's the truth. The clinical record is not the place to run an experiment. So back to the room at MIT, what Professor Cohen walked us through and what I took away from it was simple but serious. The moment you point AI at the clinical record, the diagnosis, the patient chart, the visit notes, you are standing on legal ground that most private practices are not ready for. So think about it. Your chart is not just the notes.

Legal Risk And AI Hallucinations

SPEAKER_00 2:32

It's a legal document. It is protected patient information. And if an AI tool puts something in there that is wrong or it leaks the data, the liability does not land on the software company, but it lands on you. The real risk here is the part people do not want to hear. These AI tools make mistakes with total confidence. It's called hallucination, right? They can invent a detail that never happened in the visit. I mean think about it, it's scary. They can drop something that actually did not happen. And we have a fancy technical word for it. In a plain English, the AI sometimes just make things up. And it sounds completely sure of itself. So it's very easy to get disguised and then you might not notice it. Now, take that made up detail and drop it into legal medical record. That is not a typo you fix later. That could be a malpractice exposure, you know, HIPAA exposure. That is potentially a lawsuit, and it is already happening to practice out there. Here's a screenshot of Professor Cohen's MIT slides, and I want to show you something because this is not just my opinion. I took this photo myself sitting in that room at MIT, and I want to give him full credit here. Professor Cohen, along with his colleagues Price and Girk, published this one in JAMA, a paper called Potential Liability for Physicians Using Artificial Intelligence. So now I'm not going to walk you through this whole eight of these scenarios. Okay, this is not the point of this video. But I want you to be aware that this framework exists because of what it shows. Well, these researchers mapped out who is actually liable when AI is involved in a patient care. And here's the part I want you to take away. The liability does not follow the AI, it follows the physician. So that means it follows you. And whether or not you stuck to the standard care. The software company is not standing next to you if something goes wrong. You are. And that right there is why for now I keep AI off my charts. Now I'm not saying this is impossible. If you really want to run an AI scribe, the experts in the room were very clear that there is checklist you have to clear first before AI touches a single patient note. So number one, rock solid rock solid patient consent process. Your patient have to actually understand and agree that an AI2

The Checklist Before Any Scribe

SPEAKER_00 5:23

is sitting on their visit. Number two, full number two, full HIPAA compliance, top to bottom, on how the data gets stored and handled. Number three, signed BAA with your vendor. So BAA just stands for Business Associate Agreement in plain English. And it is a legal contract that binds that company to protect. So here's the honest decision I made for practice my size. The upside of an AI scribe is real, it is convenient in time and all, but the downside privacy breach made of note potentially in a legal record, a lawsuit potentially. I mean downside is a bit catastrophic. So when the upside is convenience and downside is dangerous, the math is easy. So I keep AI completely off the clinical or diagnostic side of my charts for now. But I want you to hear that phrase for now. This is not me being afraid of AI. I just told you I run six companies on it. This is about timing and about where the risk actually sits today. Does that make sense guys? So here's the good news, and this is the part I really want you to walk away with. Clinical chart may be 10% of where AI can help your practice as of yet. The other ninety percent is wide open and it is completely safe. You know, the front desk, uh scheduling, phones, the same patient questions, uh your staff answers 50 times a day, your billing and your financial operations, intake, uh none of that has to touch protected patient information in a

Safe Places To Use AI Today

SPEAKER_00 7:03

risky way. And all of it is bleeding time and money right now. So there are AI assistants and agents and helpers private practices can plug in today on the safe side. Well, on the admin side, a system that answers your steps repeat questions and never gets tired. On the operational side, something that catches the calls, your front desk is missing. On the financial side, tools that tighten up the money quietly leaking out of your practice. So this is the stuff that gives your time back without putting your license on it. So my suggestion is to start AI where it is safe, not where it is risky. So that is why I'll not let AI near my patient charts yet, and why you probably should not either. Well, not because AI is bad, but because you start where it is safe. Now, if you're sitting there thinking, okay, TJ, I get it, but where exactly do I start? Well, that is the next video. I'm going to show you the three places to put AI to work in your practice first. So make sure you are subscribed so you do not miss this. It is free and it helps the channel grow. And if you want to go deeper, I am putting together something practical for the safe side of all of this. So

Where To Start Next And Subscribe

SPEAKER_00 8:27

keep an eye on the description below. This is Dr. TJan. I'll see you in the next one.