How I use AI
The Parent’s Question pillar and parts of the Evidence Desk pillar use AI as a working tool. Here’s what that means, so you can judge the work fairly.
What AI does here
- Literature pulls. I use AI assistants to surface research papers, clinical guidelines, and systematic reviews on a given question faster than I could on my own. I then read the primary sources myself. Anything cited has been checked by me, not by the model.
- Drafting help. Early drafts sometimes go through an AI assistant for structure and plain-language editing. The final words are mine.
- Original experiments. Evidence Desk pieces explicitly test AI tools (e.g. “which model estimates carbs best?”). In those pieces the AI is the subject, not the method.
What AI does not do here
- It does not write content that’s published under my byline without me reviewing every word.
- It does not decide what’s true. I don’t rely on model summaries of research. If I cite a study, I’ve read at least its abstract and methods — often the full text.
- It does not make medical judgments. Nothing on this site is medical advice. It’s one parent’s read of the evidence plus experience.
Why this matters
A lot of health content on the open web is now AI-generated slop with confident-sounding mistakes. The rigorous use of AI as a research assistant is not the same thing. The point of being explicit is so you can calibrate trust accordingly.
If you ever spot something on the site that reads like it was pasted in without review, please tell me. I’d rather correct it than keep it live.