Transcription
I need to tell you about something the VA just quietly disclosed, and I don't think most veterans have any idea this is even happening right now. The VA is now running 367 artificial intelligence systems across the VA department. 367. And 250 of those are classified as high impact, meaning they directly affect your health care, your benefits processing, and your records, and your claims.
AI is now listening to your doctor's appointments. AI is now pre-filling information on your toxic exposure claims. AI is now helping decide what goes into your file. This isn't coming. This is already here. It's already running right now. And depending on how you look at it, this is either the best thing to happen to the VA in decades, or it's something that should terrify you. So, let me break down what's actually happening, what's good about it, what's dangerous about it, and exactly what this means for your claims.
So, this information comes from the VA's 2025 AI inventory report, which was just released and covered by the Military Times. So, here's what we know. 367 total AI systems operating across the VA. 215 of those classified as high impact, meaning they touch critical functions like health care decisions, benefits processing, records management, and veteran services. Over 50,000 VA employees are now using commercial AI products like Microsoft Copilot, Grammarly, GitHub Copilot for tasks like meeting transcription, document drafting, record summarization, and information retrieval. The VA's chatbot on va.gov, that's also AI now, classified as high impact. The system that listens to your conversations with your doctor and writes clinical notes, that's AI now. It's called Ambient AI Scribe. The system that pre-populates your toxic exposure memoranda on the Pact Act claims or for Pact Act claims, that's also AI. It's called Terra Memorandum Automation. This is massive and the VA has been rolling it out without most veterans having any idea.
I am Dr. Marshall Barr, an actual doctor, an Army Ranger, a special operations combat medic, and a former C&P examiner. I've been inside the VA system. I've conducted these exams. I've seen how decisions get made and when I saw this report, I had two simultaneous reactions. First, finally, some efficiency. Maybe claims will actually move faster. Second, wait, who's checking this? Where's the accountability? What happens when the AI gets it wrong? Who's accountable? Because AI, in the end, it's just a tool. And like any tool, it can help you or it can also hurt you depending on how it's used and who's watching. So, let me show you both sides of the argument.
So, let me start with the good because there is benefit here. There is good here. I'm not going to fear-monger. I'm going to give you the full picture. So, let's talk about faster claims processing. The VA's Terra Memorandum Automation system helps claims processors complete toxic exposures memoranda, those, you know, those documents required by Pact Act claims, by automatically pre-populating answers using information already in your records. According to the VA, the system has a 98% accuracy rate. But we know how the VA rolls. It's been used for over 180,000 forms and it saved an estimated 54,000 work hours. That's real, and that is meaningful impact. That's 54,000 hours that claims processors can now spend on other parts of your claim instead of manually just filling out a form. If this helps get Pact Act claims processed faster, and we know the backlog has been brutal, that's a win for veterans.
There's also a benefit that is in the form of less administrative burden on providers. The Ambient AI Scribe system listens to conversations between clinicians and their patients during appointments, and then automatically generates clinical notes. So, what does this mean for you? Well, it means your VA doctor can actually look at you during your appointment instead of typing on a computer. Well, in theory at least. It means they can spend more time in theory focused on your care instead of just documentation. So, anyone who's had a VA appointment where the medical provider spent more time typing than talking knows this is a real problem. Now, if AI can fix that, that's fantastic.
Another potential benefit is now better access to information. The VA.gov chatbot helps veterans find information, complete tasks, and access support services. Now, is this a perfect system? Probably not. But, if it helps veterans navigate the system without sitting on hold for 3 hours, I mean, I'd call that progress.
Another potential benefit is diagnostic support. The VA is using AI for imaging analysis, diagnostic workflows, and disease risk identification. AI can catch patterns that humans sometimes miss. It can flag abnormalities on scans. It can help identify veterans at risk for certain conditions. If used correctly with human oversight, and if it's done this way, it could save lives. So, yes, there's real upside here. I'm not anti-AI. I am pro accountability, though. And that's where my concerns start.
So, here's what worries me about this report and what you should worry about, all right? So, concern number one, the 98% accuracy problem. Listen, the TERA system that pre-populates toxic exposure memorandum has a proposed, they're saying a 98% accuracy rate. I mean, that sounds great. 98%? That's an A+ in school, right? So, let's do the math. If that system was used on 181,000 forms and it has a 2% error rate, that is still potentially 3,600 forms with incorrect information pre-populated. 3,600 veterans whose claims might have wrong information pulled from their records. Information that a claims processor might not catch because the AI already filled it in. 98% accuracy is impressive for AI, but when it's your claim, your benefits on the line, your money, you want 100% accuracy and you're not going to get that with the current system.
Now, concern number two, variable oversight and documentation. The Government Accountability Office, the GAO, reviewed this report and raised some valid concerns. Their quote was, "As AI inputs and operations are not always visible, a central tenant to oversight it is show me, don't just tell me." Well, what the hell does that mean? It means for some of these AI systems, there's good documentation, testing results, and performance metrics. For others, though, the report just describes intended functions without showing actual outcomes. The GAO is essentially saying, "We can't verify that some of these systems are actually working correctly because there's not enough documentation to review yet." Man, that's a problem. When AI is making high-impact decisions about veterans' health care and benefits, trust us, bro, isn't good enough.
Now, concern number three, who's checking the AI? The Veterans of Foreign Wars weighed in on this report. You know, their deputy director said something I think this is exactly right. AI needs to be a tool, not an employee. And then, we need to make sure human eyes are on it afterwards. That's the question. When AI pre-populates your claim information, is a human then actually reviewing it, or are they just clicking approve because the AI already did the work for them? When AI generates your clinical notes from your doctor's appointment, is your doctor then rereading those notes to make sure there's no errors and that they're accurate, or are they just trusting the AI got it right the first time? The VA says all high-impact systems are continuously monitored, and any system that cannot be brought into compliance is then discontinued. But, the report also shows that some AI systems have already been retired, and doesn't explain the reason why. Were they not working? Were they making erroneous errors? We don't actually know.
Now, concern number four, AI and benefits decisions. And here's what keeps me up at night. AI is now involved in your benefits processing. It's pulling information from your records, it's generating documents, it's supporting the decisions that determine your rating. So, what happens when the AI misses something? What happens when it pulls the wrong service dates? What happens when it incorrectly interprets your records? Well, you have to catch it. You have to appeal it. You have to fight it, correct, you know, you have to fight to correct something that a computer got wrong. And that is a human thing, a human error, and that a human might not have caught because they trusted the computer. The burden still falls on you then, but now you're fighting against decisions that were partly made by systems you can't see or really truly understand.
So, what does this actually mean for you? For your claims, for your health care, for your benefits? Well, it means your documentation matters more than ever. If AI is pre-populating information from your records, when what's in your records is critical, errors in your service records, medical records, or previous claims could be picked up and repeated by AI systems automatically. One wrong date, one incorrect notation could propagate through your entire system. This means you need to review your records and make sure your service treatment records are accurate. Make sure your VA medical records reflect what you've actually told your doctors, and make sure your claim history is correct because AI is pulling from those records, and it's only as good as the data it has to pull from.
Another thing is you need to start to read everything the VA sends you. So, when you get decision letters, proposed ratings, or any documentation from the VA, please just read it carefully. Don't assume it's correct because the system generated it. Look for errors. Look for information that doesn't match your actual situation. If something is wrong, correct it immediately. File for correction of records. Challenge inaccurate information in appeals.
Another thing you're probably going to have to start doing is where your personal statements are going to need to become more valuable. So, here's something AI can't do. It can't tell your story. AI can pull dates from records. It can summarize medical notes. It can pre-populate forms, but it cannot explain what you experienced. It can't describe how your symptoms affect your daily life. It can't capture the nuance of your condition. Your personal statement, your buddy statements, your spousal letters, these are all human documents that provide context that AI just can't generate. So, make them detailed. Make them specific. Make them impossible to ignore.
Another thing that's going to happen is human appeal rights are still in existence, right? And this is important. No matter how much AI is involved in this process, you still have the right to appeal. So, you still have the right to request a hearing with a human. You still have a right to submit additional evidence in a supplemental claim. You still have the right to challenge any decision the VA makes. AI might be involved in the process, but it doesn't have the final word. You still do.
Another thing is to ask questions at your appointments now. If you have a VA medical appointment and you notice the doctor isn't typing, if they mention that notes are being generated automatically, ask about it. Ask, "Can you read me back what the AI documented about our conversation?" Ask, "Is everything accurate?" You have the right to know what's being put in your medical records. And you have the right to correct it if it is wrong.
Now, this isn't just about the VA. AI is transforming literally everything. The world around us, every government agency, every healthcare system, every industry. This is the world that we're moving into. And honestly, for an organization like the VA, which has been plagued by backlogs, delays, and inefficiencies for decades, AI could really be a game-changer. It could help veterans get faster decisions, better care, more accurate processing, but only if it's done right. Only if there's accountability, only if humans are still checking the work, and only if veterans are informed about how these systems are being used. Cuz right now, we're in the early stages. The VA is deploying these systems faster than most veterans realize, and the oversight is, well, let's call it variable.
The VFW is right. AI needs to just be a tool, not an employee. And the GAO is right. Show me, don't just tell me. We should demand transparency. We should demand documentation. We should demand that when AI is involved in decisions about our benefits, we know about it, and that we can actually verify that it's accurate.
So, here's what I want you to do. Multi-steps here. So, step one is get copies of your records, your service treatment records, your VA medical records, your claims file, and review them for accuracy, because AI is pulling from those, and errors will propagate. Step two is read every document the VA sends you. Don't just skim. Don't assume anything. Look for errors, inconsistencies, information that doesn't match your situation. Step three is write detailed personal statements, your story, your symptoms, your daily limitations, in your own words. This is something AI can't replicate or replace. Then step four, ask questions at your VA appointments. If notes are being generated by AI, ask to verify what's being documented and make sure it's accurate. Step five, if you find errors, appeal immediately. Don't let incorrect AI-generated garbage sit unchallenged in your file.
Now, the VA is running 367 AI systems, 250 of which are considered high impact, which affect your health care, your claims, and your benefits. This is both an opportunity and a risk. The opportunity, like I said, faster processing, less administrative burden, better efficiency. But, the risk is also very real. Errors that could be propagated, decisions you can't see, accountability that's kind of unclear. Clear as mud, as I would say. Your job is to stay informed, to stay vigilant, to make sure your records are accurate, that your statements are detailed, and your voice is replaced by an algorithm.
Now, if you need help navigating into this, that's exactly what I do and we do every day at X-Terra Health. We are physician-led. I am a medical doctor and a former C&P examiner. I understand how the VA system works, including how documentation becomes evidence, and how evidence drives your rating results. We offer Nexus letters written by medical providers who know what the VA requires. Not templated garbage letters, but individualized medical opinions that stand up to scrutiny, whether that scrutiny comes from a human or even an AI system. We also offer claims road map services, where we review your records and your claim strategy. Then, we identify errors, gaps, and opportunities you may never have identified. We make sure what's in your file actually reflects your situation. And, we also offer C&P exam preps, where we prepare you to describe your symptoms accurately and thoroughly, because what you say in that C&P exam, and what gets documented, matters more than ever. So, if you're interested in any of these services, the links for the the the links for these services are in the description below.
Now, I am Dr. Marshall Barr, an actual doctor, an Army Ranger, a former C&P examiner. AI is here, and it's not going away, but neither are your rights. Just stay informed, stay vigilant, and don't let a computer make decisions about your benefits without your input. So, subscribe if you want to know what the VA isn't telling you. I'll see you in the next video, and until next time, stay mission ready.