A Quick Note About How We Do and Don't Use AI in This Newsletter
With AI slop proliferating around us, we want to be clear about our commitment to being human-produced.
Let’s talk about AI for a moment and its use in our newsletter. We’ll be brief!
Every article, analysis, and opinion in this newsletter comes from people who do regulatory, quality, or clinical work for a living. Our contributors are consultants, former FDA personnel, quality system specialists, and other industry practitioners who have spent years inside the environments they write about.
When we break down a warning letter and pull lessons from it, that insight has come from one of our senior consultants on our contributor team.
We don’t accept AI-generated anything from our contributors. That’s a firm policy that, in full transparency, we’ve had to start enforcing, given how “good” LLMs are getting at generating content. We want this newsletter to be a refreshingly human source of information.
So let’s talk about it.
Where AI shows up in our process
We’re fully committed to keeping our content human and experience-sourced because we know it’s what you’re here for. But we’re not going to be digital Luddites and deny ourselves tools that support that commitment.
Some of the datasets we work with, things like warning letter trends, inspection citation patterns, and adverse event signals, are large enough that manual analysis would take days or weeks or wouldn’t happen at all. We use the most advanced models to let us ingest that data and identify patterns. But the interpretation, context, and the editorial judgment about what those patterns actually mean still come from our team.
Also, much of what we publish here starts with interviews with subject matter experts in our consulting network. We meet on Zoom and have conversations to extract insights firsthand. These conversations are often dense; they jump around, and the best stuff is often buried in an aside fifteen minutes in. We use AI to extract raw insights from those transcripts and organize them so we can write from them more efficiently and make sure nothing gets lost. The thinking and the opinions are the SME’s, but AI helps us make sure we capture all of it before we start writing. It would take hours to sift through transcripts manually, and honestly, a human misses more than a machine does at this point.
We also use AI to catch typos, flag inconsistencies and inaccuracies, and check whether a paragraph reads clearly before publication. Neither we nor any tool is perfect, so we do ask for an ounce of grace when things are missed. We’re a very small team with ambitious publishing goals. The ideas, arguments, and conclusions are already on the page before any AI tool touches them.
That’s about it when it comes to where AI and LLMs show up in our workflow. To be clear, we don’t use AI to generate insights or analysis from scratch, and we don’t feed it a topic and publish what comes back. Anyone with an internet connection can do that at this point.
What we don’t do
Just to put a finer point on this, we don’t generate insights from whole cloth, and we don’t allow AI models to train on confidential or sensitive client information. Our contributors work with regulated companies, and the trust those relationships require extends to every tool we use.
We also don’t use AI to manufacture perspectives we don’t hold, to pad word counts, or to produce content faster than our contributors can think through it. Speed isn’t the point here.
At the risk of being repetitive:
We do not use AI to generate insights or analysis from “nothing.” We use it only to extract usable insights from raw contributor input, typically in the form of an unstructured transcript.
We do not use AI to actually draft what we write. In other words, we don’t start with a blank page and have AI fill it in. We use it on the edges to help us refine and clarify something a human contributor wrote or catch typos and other proofreading mistakes. (If you see a turn of phrase common in AI writing, it could be a bit of assisted writing, but sometimes it’s actually us writing that way!)
We will use AI to check the accuracy of facts we present. This is a genuinely useful task for an LLM: tracking down and confirming or refuting factual claims against some authoritative source. It can be a great fact-checker if it’s used correctly.
This matters more here than it might elsewhere
There’s a practical reason this distinction matters here and why we want to be super clear about it.
In any business content, a wrong detail is embarrassing. In the life sciences, a wrong detail about a regulatory pathway, a misread of a guidance document, or a sloppy characterization of an enforcement trend can lead someone to make a bad decision about a product that affects patient safety. The cost of getting it wrong in our space is actually operational and sometimes clinical.
This is also a field where experience-based judgment counts for a lot. Knowing what regulators are likely to focus on during an inspection, or how the FDA tends to interpret a given question, isn't knowledge that tends to live in LLM datasets you can conjure with a prompt. It lives in the brains of the people who’ve actually been in those rooms.
That’s what we’re trying to preserve in this newsletter. Not because AI is “bad,” but because the content that actually helps our readers requires something AI can’t provide on its own in the same fidelity or depth.
Why we care about this enough to write about it
We started this newsletter because we noticed the same thing many of you have: the amount of life sciences “content” online keeps growing, and most of it is “thin” to put it nicely. There’s a lot of generic advice that applies to everything and helps with nothing.
So when someone asks whether we use AI, we take it as a reasonable question from a reader trying to figure out what they can trust. We hope this answers it. If you have specific questions about this or anything in general, just email us:
thefdagroup@substack.com
Who is The FDA Group?
The FDA Group helps life science organizations rapidly access the industry's best consultants, contractors, and candidates. Our resources assist in every stage of the product lifecycle, from clinical development to commercialization, with a focus on Quality Assurance, Regulatory Affairs, and Clinical Operations.
With thousands of resources worldwide, hundreds of whom are former FDA, we meet your precise resourcing needs through a fast, convenient talent selection process supported by a Total Quality Guarantee. Learn more and schedule a call with us to see if we’re a fit to help you access specialized professionals and execute your projects on time and on budget.