Subscribe to The Podcast by KevinMD. Watch on YouTube. Catch up on old episodes!
One AI-hallucinated citation on cross-examination, and the expert witness career you built is over. It is already happening. Tracy Liberatore, a former physician assistant turned attorney and founder of the National Expert Academy, walks through the real court cases where clinical experts leaned on generative AI and watched their reports, and their reputations, get thrown out. This episode is based on her article “Expert witness credibility is destroyed by AI opinions,” published on KevinMD. You will hear why AI hallucinated citations are ending careers in medical-legal work, why one expert was allowed to keep AI in his workflow because he could account for every prompt, what responsible AI use actually looks like for clinicians writing expert reports, and the brain flip clinicians have to make to defend a process rather than a conclusion. If you do expert witness work, or are thinking about it, this conversation names the line you cannot cross.
Partner with me on the KevinMD platform. With over three million monthly readers and half a million social media followers, I give you direct access to the doctors and patients who matter most. Whether you need a sponsored article, email campaign, video interview, or a spot right here on the podcast, I offer the trusted space your brand deserves to be heard. Let’s work together to tell your story.
PARTNER WITH KEVINMD → https://kevinmd.com/influencer
SUBSCRIBE TO THE PODCAST → https://www.kevinmd.com/podcast
RECOMMENDED BY KEVINMD → https://www.kevinmd.com/recommended
Transcript
Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast. Today we welcome Tracy Liberatore. She’s the founder of the National Expert Academy. Today’s KevinMD article is “Expert Witness Credibility Is Destroyed by AI Opinions.” Tracy, welcome to the show.
Tracy Liberatore: Thank you. Appreciate it.
Kevin Pho: All right. Let’s briefly introduce yourself and then talk about why you decided to share this particular article on KevinMD.
Tracy Liberatore: Sure. So I am, as you said, Tracy Liberatore. I was a physician assistant for about 10 years and went back to law school, and I’ve been running Med-Legal Pro for about 10 years before launching the National Expert Academy the past month. And I chose this particular article because AI is just such a hot topic these days. It seems everywhere you go, somebody’s talking about it in some capacity.
And because I have integrated it now into some of my trainings, and I feel like the best use is something, it’s just a very interesting topic, I think, for people. And I think a lot of people are afraid now of losing their jobs because of AI. And so I wanted to just talk about how that might impact the expert witness field.
Kevin Pho: All right, so tell us about that for those that didn’t get just to read your article. How is AI affecting the expert witness field?
Tracy Liberatore: Yeah, so I don’t actually have my article in front of me, but I was just at a conference yesterday, and the legal field is constantly talking about AI use in court and in trial, and even with lawyers using it and the AI hallucinating and citing cases that are not accurate.
And as far as medical experts are concerned, we’ve got cases since 2025. Starting with Kohls v. Ellison, where Jeff Hancock was retained as an AI expert to defend Minnesota’s anti-deepfake law, and he used a GPT to help draft his declaration, and it fabricated citations. So there were non-existent academic articles that were cited. And the irony kind of writes itself that the AI expert was taken down by AI.
There’s Concord Music Group versus Anthropic, where the data scientist employed by Anthropic submitted a report that AI had hallucinated citations. So this is sort of the theme of all of this AI stuff.
And there was an interesting case that I thought, if I’m pronouncing it right, Iovino v. Harbor Freight, where the court actually refused to exclude an expert who used some AI, but he was only using it to double-check his findings. And so he was able to account for what he had used it for, and so the courts had allowed it in.
The biggest thing I’m seeing is really, if you didn’t write it and if you can’t talk about where you got your information from, you’re probably going to be in trouble for it.
Kevin Pho: So, as it relates to expert witness work, why is AI use such a high-stakes issue in that realm?
Tracy Liberatore: Because an expert witness is on the stand, we’re talking about these things based on their clinical judgment, the standards of care and clinical practice, the things they’re seeing with patients. And so, number one, the hallucinations are an issue, but if you can’t speak to your expertise, that’s really what you’re hired for. And there’s that human element, the judgment piece that’s really important for speaking to causation and the issues that are important for the cases.
Kevin Pho: So are we seeing more and more physicians and clinicians using AI for their expert witness testimony?
Tracy Liberatore: Unfortunately, it seems like that is the case with some of these cases that are coming out. I think clinicians are really becoming more aware of it. Certainly judges are becoming more aware of it. They’re talking about this a lot.
But again, I think everything comes down to the tool and how are you going to use the tool and what’s appropriate. So the appropriate use, I think, for AI is if you’re going to have it help you with training, learning something, helping you get organized. But at the end of the day, that opinion needs to be yours. It’s your license, it’s your clinical judgment. You’re signing that report, and if you can’t account for the stuff in that report, how the heck are you going to get on the stand and defend it?
Kevin Pho: So tell us what goes into writing an expert testimony. Just from a legal standpoint, what do physicians need to know if they’re interested in doing this work?
Tracy Liberatore: Yeah, so that’s another big topic. I always tell everybody, it took me a year of law school to learn how to write legally. I remember the first 60 that I got and my stomach dropped out trying to figure out how to do this, and it’s just a different way for a clinician’s brain to have to think.
I was trained doing SOAP notes, subjective, objective, assessment, and plan. And so we only really care about, well, what’s wrong with the patient? What are we going to do to treat them? And we’re passing this information along to the next clinician in line that might have to come in and treat them.
Well, when you’re coming in to write an expert report, everything that you think about that goes through your brain to get to your conclusion has to make it to the paper. And a lot of times what I see when people are, especially new people, when they come in, they use weak language, like, “Well, maybe this would’ve changed the outcome.” And they just don’t get all of the stuff that runs through their head on a paper. Why did this person suffer this? Was it something they did do, or that they omitted to do? And then how did this change the outcome for this person? If they would’ve done what they were supposed to do, what would’ve been different for them?
And so the damages piece and the causation, leaning to the damages and talking about all of that, the analysis and how does it apply to the standard of care, which is your rule statement. And it’s a brain flip, as I’ve called it in my other article.
Kevin Pho: Yeah. So I could see the temptation of physicians who aren’t comfortable writing legally turning to AI to help them write. I know you offer a course in this, but is that the only way where physicians can learn to write credible expert witness reports? Is there any other type of training they should be looking for?
Tracy Liberatore: Yeah, well, it’s understanding the standards. I mean, you’ve got Daubert standards that everybody talks about when it comes into this stuff. But really, I’ve worked one-on-one with people for 10 years and I still have some people that come, I still look at their reports and they still sort of make the same mistakes. Certain things that you don’t want to say. You don’t want to be advocating for the patient. You don’t want to say the word negligence in your report. I mean, this is what we’re all talking about, but that’s a jury question. So putting that word in your report, you need to talk about the standard of care, which is what you’re commenting on.
If there’s other courses, there’s other people that have courses out there too. I’m not the only one, if that’s your question, but I of course think my course is the best. And I have created what I believe is the first use of an AI guide within a course. So when you take my course, I’ve created the practical exams that you could submit and immediately get an attorney’s perspective feedback. My 10 years of working in this medical-legal industry is put into this, so it’s trained for each module to give you the feedback immediately. You don’t have to wait for somebody to give you some feedback or do a self-diagnosis. That I believe is the first of its kind. I could be wrong.
Kevin Pho: Now, is this the only way for physicians to learn to be expert witnesses? There’s no academic setting, for instance. Just these third-party courses?
Tracy Liberatore: I think so. I don’t know of any academic setting where you can actually go, outside of going to law school, though. I mean, you can always take a tort class or something like that. That’s the route I went. I personally thought I wanted to be a legal nurse consultant, but back when I wanted to do it, I couldn’t get into LNC school because I wasn’t a nurse. They wouldn’t accept me. So that’s why I went to law school, to get the education.
But I don’t think for being an expert, you don’t need law school. You just really need to understand tort law so that you can understand what you need for the legal side of things. And then really, I think getting somebody that has a perspective such as myself, that can understand the medicine and understand the law, so look at your report and give you one-on-one feedback how you can improve that particular report. That’s how I do it. We go back and forth. I think that’s probably the fastest way.
Kevin Pho: Now for physicians who are writing expert witness reports, what would responsible use of AI look like?
Tracy Liberatore: In my opinion, again, using it, I think in a training setting like that so that you can get that feedback on your exact writings. But also if you need it to help you with research or to get a rule offline or to organize your thoughts, I think that would be OK. As long as you can account for what are you using it for, what are your prompts that you’re using it for, and double-check your citations.
Don’t put something in there. Use it as a shortcut. It’s hard sometimes for people to find these cases and stuff, but if you’re going to cite something, do your homework, go read the actual case. It won’t take you that long to do, and be responsible about using it.
Kevin Pho: So what are the repercussions if a physician gets caught, oh my goodness, using AI inappropriately? What happened to some of these cases that we talked about?
Tracy Liberatore: Yeah, so first of all, your report’s going to get thrown out of the case. So you’re going to hurt the attorney, you’re going to hurt the case, and your credibility goes right out the window. I mean, credibility as an expert is everything. It is everything. And if you lose that, you just won’t have a business anymore. You won’t be able to serve as an expert. Unfortunately, nobody’s going to hire you again.
Kevin Pho: But in terms of legal ramifications, are there any legal trouble that these expert witnesses can get in by fabricating reports or fabricating citations using AI?
Tracy Liberatore: I would imagine there would have to be, and I didn’t do research on that, so I couldn’t cite a case for you on that, but I can’t imagine that there wouldn’t be. I mean, if you’re putting false information in an expert report that you’re testifying under oath in a legal proceeding, and then if you’re up on the stand in your case, it gets thrown out because of your using AI. If your attorney loses that case, I can’t imagine, if you’re the reason for that, that you wouldn’t be sued for whatever they might have been able to recover for the case. I mean, you’re harming the client, you’re harming the attorney. It’s just not a good spot to be in, for sure.
Kevin Pho: Now, for those physicians who may be listening to us today, tell us the type of characteristics or traits they may have if they were to consider, not a career change, but adding expert witness work to their plate.
Tracy Liberatore: Yeah, so I talk to a lot of attorneys on my podcast, Statutes and Stethoscopes, and my main thread through that is asking, what are you looking for when you’re hiring an expert? And I think that’s the question you’re asking me. And for a lot of them, it’s really being candid with the case, being honest with things. They do not want you to be biased. They do not want you to advocate for their client. They want you to look at the medical record and give them your honest opinion on, do they have a case or not. They want to know the strengths, they want to know the weaknesses.
The other complaints that I’ve heard over the 10 years that I’ve been doing this as experts: take too long to get back to me. I don’t know what’s going on with my case. Communication. It’s just a huge thing. They don’t want to be surprised with a major invoice. If you’ve got 10 hours in, on occasion, you might want to check in with them if you think it’s going to take 20. Expectation management is huge, and that’s, again, a big part of what we do with my company, with Med-Legal Pro.
Kevin Pho: We’re talking to Tracy Liberatore. She’s the founder of the National Expert Academy. Today’s KevinMD article is “Expert Witness Credibility Is Destroyed by AI Opinions.” Tracy, let’s end with some take-home messages they want to leave with the KevinMD audience.
Tracy Liberatore: Sure. So I guess what I would leave you with is, every one of the cases that we discussed today, the expert really just couldn’t account for their own process. They could not separate what they had concluded from what the machine had produced. And that’s really the collapsing point. And it’s really what the clinicians are at risk for the most, because the clinical training doesn’t teach you to defend a process. It teaches you to defend a conclusion. And so if you’re going to get into this type of work, you sort of have to flip that in your brain. Your report is your artifact, and the process is what gets cross-examined.
Kevin Pho: Tracy, thank you so much for sharing your perspective and insight. Thanks again for coming on the show.
Tracy Liberatore: Yep. Thank you.


















