Before You Decide: A Vaccine Research Tool
How to use AI to find the evidence that exists — and the evidence that doesn’t
Two separate conversations led to this.
A friend had started asking ChatGPT questions about vaccines. Not leading questions — just direct ones, following wherever the answers led. I watched his thinking develop in real time through the questions he asked:
How does a child catch diphtheria?
Of the countries that do not vaccinate for it, what are their numbers of cases?
How many cases of diphtheria have been found in Australia from 2020 onward?
What is the population of Nigeria?
How many children born in Nigeria are vaccinated for diphtheria?
Assuming an average of 500 deaths per annum, what percentage is that of the number of unvaccinated children?
You can see what he was doing. He wasn’t trying to catch the AI in anything. He was trying to understand the actual risk — not the risk as presented in public health messaging, but the risk calculated from the numbers.
What came back surprised him. Not misinformation or conspiracy theories — the opposite. Specific numbers that didn’t match the urgency he’d been told to feel. The AI, trained on mainstream sources, was telling him things his doctor never had. Not because ChatGPT was “anti-vaccine,” but because he’d asked questions precise enough to surface what the data actually shows.
Independently, Amelia — one of my earliest mentors in this space — proposed something similar. What if we created a set of questions that parents could use with AI? Not to generate scary stories, but to access the primary sources: package inserts, clinical trial documents, regulatory filings. The evidence that exists, and the evidence that doesn’t.
Those two conversations produced what I’m sharing today: a document called Before You Decide: A Vaccine Research Protocol.
What This Is
The PDF embedded below is a research tool. It contains prompts — questions designed to surface specific information — that parents can copy and paste into any AI assistant (ChatGPT, Claude, Perplexity, or others) to find documented evidence about vaccines and related interventions.
The prompts are built around a framework I’ve written about before: the Three-Legged Stool. Before any medical intervention, three questions must be answered:
Is it necessary? What’s the actual risk? What are the alternatives?
Is it safe? What are the documented harms? Were they honestly assessed?
Does it work? In outcomes that matter, not just surrogate endpoints?
If any leg fails, the stool cannot stand. The PDF teaches parents how to check each leg — not by telling them what to conclude, but by giving them the questions that surface the evidence.
Why AI?
AI assistants are trained on medical literature, regulatory documents, and scientific studies. They know what’s in the package inserts. They know what the clinical trials measured and for how long. They know which studies exist and which don’t.
The problem is that AI is also trained to defer to institutional consensus. Ask a vague question, get a reassuring answer. Ask “are vaccines safe?” and you’ll receive boilerplate.
But ask a precise question — What was the control group in the clinical trials? Was it saline or another active substance? — and the AI has to engage with what the documents actually say.
These prompts are designed to ask precise questions. They instruct the AI to present absolute numbers rather than reassuring language. To acknowledge when evidence doesn’t exist. To apply the same evidentiary standards to official claims and critical claims.
The AI isn’t the enemy. It’s a tool that reflects what it was trained on. Ask the right questions, and it will tell you what the data actually shows — including when the data isn’t there.
What Parents Discover
Here’s a real example. When you use the PDF’s prompt asking about hepatitis B vaccine ingredients, ChatGPT responds with a detailed breakdown — every ingredient, its purpose, and then this:
For aluminium adjuvants:
“I did not find human studies where aluminum hydroxide/aluminum hydroxyphosphate sulfate is injected by itself into infants with subsequent long-term follow-up and organ distribution measurements.”
For formaldehyde:
“I did not find studies where formaldehyde is injected alone into infants specifically to establish a safety profile at vaccine-relevant doses.”
For polysorbate:
“I did not find human infant studies that inject polysorbate 20 or 80 alone intramuscularly and then measure pharmacokinetics or long-term outcomes.”
For borax (an excipient in one of the Australian hepatitis B vaccines):
“I did not find studies where sodium borate is injected intramuscularly into infants to establish infant injection-specific pharmacokinetics or toxicity.”
Four ingredients. Four admissions that no injection-specific infant safety studies exist.
This isn’t ChatGPT being “anti-vaccine.” This is ChatGPT accurately reporting the state of the evidence when asked a direct question. The safety leg — for these ingredients, in this population, via this route — has not been tested. It has been assumed.
That’s the pattern parents discover when they use these prompts. Plain-language admissions about what has and hasn’t been studied, coming from an AI trained on mainstream medical literature.
How to Use It
The PDF has two tiers:
Tier 1: Essential Questions — Five core prompts that address all three legs of the stool for any vaccine. If you’re short on time, start with Question 2 (How was safety tested?). What you learn about placebo controls will reframe everything else.
Tier 2: Deeper Investigation — Detailed prompts for specific concerns (aluminium, autism, timing) and specific vaccines (Hep B, DTaP, MMR, HPV, flu, and vitamin K).
Copy a prompt. Paste it into ChatGPT, Claude, or Perplexity. Read what comes back. Follow the citations to primary sources if you want to verify.
The document also includes:
A guide to recognising non-answers from medical professionals
A printable one-page summary to bring to appointments
An explanation of the “iatrogenic cascade” — how early interventions shape which later interventions you’re offered
Sharing This With Someone You Love
The PDF is designed to be shared. It doesn’t tell parents what to decide. It gives them a way to evaluate the evidence themselves.
If you have a loved one — a pregnant friend, a new parent, a family member facing these decisions — and you want to share this without scaring them off, here’s an email you can adapt:
Subject: Something that helped me
Hey — I came across something that helped me think through some questions I’ve been sitting with about some of the early medical decisions for babies.
It’s a research tool — a set of questions you can ask AI (like ChatGPT) to get specific information from official sources. Not someone’s opinion, just a way to find the actual evidence.
I found it useful and thought of you. No pressure to read it or use it — just wanted to pass it along in case it’s helpful.
Happy to talk if you want, but no pressure either way.
[Your name]
The goal isn’t to convince anyone. It’s to give them a tool they can use when they’re ready.
The PDF
The stool either holds your weight, or it doesn’t. These questions help you check the legs before you sit.
Book: Medicalized Motherhood: From First Pill to Permanent Patient
Available as a free download. 123 interventions documented across six phases—from pre-conception capture through postpartum surveillance. Includes practical tools: birth plan template, provider interview questions, quick reference card, and a new chapter on interrupting the cascade. Download it, share it with someone facing their first prenatal appointment, their induction date, their cesarean recommendation. The cascade works because women don’t see it coming. This book makes it visible.
Support Independent Research
This work remains free because paid subscribers make it possible. If you find value here, consider joining them.
What paid subscribers get: Access to the Deep Dive Audio Library — 180+ in-depth discussions (30-50 min each) exploring the books behind these essays. New discussions added weekly. That’s 100+ hours of content for less than the price of a single audiobook.
[Upgrade to Paid – $5/month or $50/year]
Get in touch Essay ideas, stories, or expertise to share: unbekoming@outlook.com
Bitcoin: 3Q6BK8x8zjoPaXykQggzvoJxg5FiEbkb3U
Ethereum: 0x4CB0d39d8466a34609318FC1B003B745893788b3
New Biology Clinic
For those of you looking for practitioners who actually understand terrain medicine and the principles we explore here, I want to share something valuable. Dr. Tom Cowan—whose books and podcasts have shaped much of my own thinking about health—has created the New Biology Clinic, a virtual practice staffed by wellness specialists who operate from the same foundational understanding. This isn’t about symptom suppression or the conventional model. It’s about personalized guidance rooted in how living systems actually work. The clinic offers individual and family memberships that include not just private consults, but group sessions covering movement, nutrition, breathwork, biofield tuning, and more. Everything is virtual, making it accessible wherever you are. If you’ve been searching for practitioners who won’t look at you blankly when you mention structured water or the importance of the extracellular matrix, this is worth exploring. Use discount code “Unbekoming” to get $100 off the member activation fee. You can learn more and sign up at newbiologyclinic.com
25 years ago I did my research on vaccines, basic info was found on early days of internet. I found that it contained something from monkeys so I said no way. Also had friends who had healthy children which had never had any poison. I have been telling people since but to convince those who are adamant that vaccines work it is hard to get them to even consider doing any research.
My man! Dude - here's a guide I created on what to do IF a vaccine reaction does occur. Time is off the essence and these are the steps taken so that one can document everything and be able to win in court if it goes there: https://shadowbannedlibrary.com/products/what-severe-vaccine-reactions-look-like-ebook-by-unorthodoxy