Understanding AI in Healthcare: Transparency and Ethics
I have been obsessed with AI for a while now. I’ve taught workshops on it, I use it in my own practice, and I think about it probably more than is healthy for someone who also has actual clients to see.
I also think it’s plagiarism. I went on strike for it in 2023. And I will not use it to write my own content — because creativity requires humanity, and that’s not a negotiable position for me.
Both of those things are true at the same time, and if that sounds contradictory, welcome to the conversation. This is not a topic that resolves cleanly into pros and cons. But I’m going to try to give you the most honest version of where I’ve landed, because I think you deserve something more useful than “AI is amazing!” or “AI is the end of everything we hold dear.”
What AI Actually Is (and Isn’t)
Here’s the simplest way I can put it: AI is programming and algorithms that analyze enormous amounts of data to find patterns and make predictions. Think of it less like a brain and more like a very, very sophisticated autocomplete — one that’s been trained on essentially everything ever written on the internet, which is both why it’s impressive and why it’s a problem.
ChatGPT will tell you it’s “a field of computer science focused on creating systems capable of performing tasks that usually require human intelligence.” That’s ChatGPT marketing itself to you. It is not an objective summary. And that’s actually the first thing you need to understand about working with any of these tools: they are not providing you with objective results. There is always going to be some combination of bias, error, and what the industry cheerfully calls “hallucinations” — which is a delightful word for “it made something up with complete confidence.”
I have tried several different tools for clinical documentation. One of them decided, apparently based on having been trained on a blog post that mentioned a well-known tongue tie specialist, that every single one of my clients needed to fly across the country for a consultation with a doctor I never mentioned. It does this constantly. It’s annoying, but it’s fixable, and stands as a good reminder that you cannot turn your brain off just because a robot is taking notes for you.
Where AI Can Genuinely Help You
I want to be honest about this part, because I think knee-jerk skepticism can be just as unhelpful as the uncritical hype.
Clinical documentation. This is where I have seen the biggest real-world difference, for me personally and for IBCLCs I’ve talked to. Tools like [affiliate links] Scope, TwoFold (use code AnnieF for 7% off), Heidi, Freed, and Nabla listen while you work and generate a draft SOAP note from the transcript. We’re talking about 20–30 minutes of charting time back per visit. I ran an online workshop last year on AI and we had several comments about how much the AI saved time. One person said she was able to double her weekly clients after starting to use Heidi. Another said she has ADHD and the cognitive load of trying to remember everything while also being present was genuinely unsustainable.
The attention piece is what I value the most. Having the AI pay attention for me has given my attention back to my clients. I am not younger than I used to be. If I see three clients in a day and don’t write everything down immediately, I cannot reliably reconstruct the first visit by the time the third one ends. Knowing the AI has it means I can actually be in the room with families instead of mentally trying to stuff everything into my brain.
Administrative tasks. Letters of medical necessity, insurance appeal letters, referral correspondence — anything where you’re starting from a blank page and the structure is predictable. AI can generate a solid first draft that you then review and edit. You’re not staring at a blank page. That’s real.
Research and education. AI can be a useful starting point for finding search terms, getting an overview of a topic, or looking for resources you might not have known to look for. Emphasis on starting point. It will cite sources with complete confidence that it has absolutely fabricated. Verify everything before it goes near a client or a chart.
Scheduling and intake automation. Some practice management platforms are building AI features that handle reminders, pre-populate intake forms, and streamline scheduling. Less time on administrative logistics is a pro.
Where AI Will Get You Into Trouble
The corporate interests behind these tools are not your interests. OpenAI, Google, the studios — they have all demonstrated, pretty consistently, that what they care about is their stock price and their ability to do what they want with everyone else’s intellectual property. ChatGPT is training on your inputs unless you have a paid account and have explicitly turned that off. The free version of Gemini is not HIPAA compliant. Google Workspace Gemini is — but only if you’re using it inside your HIPAA-covered workspace, not through the app or at gemini.com. These distinctions matter, and the companies are not going out of their way to make them obvious.
Your client data is not theirs to train on. The clinical tools — Heidi, Freed, Nabla — have all told me they do not use client data to train their models, and they have data processing agreements in place with whatever large language models they’re using. I believe them, provisionally, because there would be significant legal exposure if they were lying. But ask the question directly: Do you use my consult data to train your AI? You deserve an answer.
Your client relationships are not a workflow problem. The whole reason families come to an IBCLC instead of just Googling is that they need a human being — someone who will actually listen to their birth story, notice that they’re scared, and respond to what’s actually happening in that room. AI can simulate empathy. It cannot provide it. I am hopeful, when I go to the doctor and they are giving me their full attention for once, that it’s because some AI is taking notes so they can be present with me. That’s the right use. The wrong use is letting the tool do the part that only you can do.
Your voice is your brand. If you’re using AI to write your blog posts, your newsletters, your social media content — your audience will eventually notice. Not because AI content is obviously bad, but because you won’t be in it. The specific things you find funny or outrageous, the way you frame something that nobody else would frame that way, the stories only you have lived — that’s what turns a casual reader into someone who buys your course and sends their colleagues your way. An AI can write competent content. It cannot write your content.
The environmental cost is real, and I don’t have a clean answer. Data centers use enormous amounts of water and electricity. There are communities living next to server farms dealing with noise levels that have caused physical harm. I think about this. My working resolution is “don’t let the perfect be the enemy of the good, and also push for accountability from the companies building this infrastructure.” That’s not satisfying, but it’s honest.
AI is a threat to our profession if we let it be. There are people who would very much like to use AI to replace high-touch lactation care with something cheaper — a chatbot, an app, a text-based support tool that a startup can scale without paying anyone a living wage. We’ve seen it happen in other sectors. The writers went on strike over it. I’m not being paranoid; I’m paying attention to trends. The fact that AI tools can help you be better at your job does not mean the people building those tools have your job’s best interests at heart.
If You’re Using AI Clinically, Your Clients Need to Know
This one is non-negotiable for me, and it connects directly to something I ask of you in your practice: tell the truth.
If you’re using any AI tools — even just for documentation — your clients have a right to know. Families are sharing intimate, vulnerable information with you. They deserve to know how that information is being handled.
I put together a Clinical AI Disclosure & Resources toolkit to make this easier. It includes a client-facing disclosure that covers how you’re using AI, why you’re using it, and how their data is protected — attorney-reviewed, and worded to actually reassure people rather than just cover you legally. It also includes prompts for [affiliate links] Scope, TwoFold (use code AnnieF for 7% off), Heidi, Freed, and Nabla, ChatGPT, and Google Gemini, with more being added as tools evolve.
If you’re already using AI in your practice, get this before your next consult. If you’re planning to start, get it first.
A note before I wrap up: I have complicated feelings about this post existing at all.
I used an AI tool to help me write it. Specifically, I trained it on my own previously published writing and used it as a drafting assistant. Think of it less like “AI wrote this” and more like “AI helped me organize and articulate thoughts I’ve already said in a dozen different places.” Every idea here is mine. The voice is mine. But I didn’t type every word from scratch, and you deserve to know that.
Here’s why I’m telling you: I just spent several paragraphs making the case that your clients have a right to know when you’re using AI. I believe it’s important to disclose my own usage. Transparency matters.
I feel strongly that writers should write. I am a writer. I will continue to be the one writing everything that goes out under my name, and I’ll be transparent when the process looks different than you might expect. That’s the only honest way I know how to do this.
The Bigger Question
Working with AI tools is a little like trying to guide a plow horse that keeps getting distracted by a noise off to the side or wanting to sniff at an interesting rock. You’re not going to get it to do exactly what you want every time. You’re going to end up with some hallucinations, some weird suggestions, some notes that send your clients to a specialist in a city they’ve never visited. You review it, you fix it, you move on.
But here’s what I keep coming back to: why did you become an IBCLC?
Not to optimize workflows. Not to scale content. Because you wanted to be in the room with families at one of the most vulnerable, most meaningful moments of their lives — and help them find their way through it.
AI is a tool. Use it for the paperwork. Keep it away from the relationship. Tell your clients when you’re using it. And whatever you do, don’t let it write the words that are supposed to sound like you.
Those words matter. They’re how the right families find you, trust you, and eventually tell other families about you. That’s not something you want to outsource.
