This week, I sat in on a high-stakes internal conversation about AI implementation. Here's what I learned.
Claude's "learning mode" invites a slower, more deliberate use of AI — one that might reshape how we think about education, agency, and the internet itself.
Untrained Data is a newsletter, journal, and conversation space for people learning to live and build with AI. Written by Richard Udell, it blends tech commentary, interviews, and personal stories from the edge of what's next.
Untrained Data is built with help from large language models — including the site itself and much of the writing. But the ideas, direction, and edits are mine.
This project is about using AI as a tool, not a replacement. I use models to speed up drafting, explore ideas, and get past blank pages. But I stay in the loop — reviewing, rewriting, and shaping every piece myself. If I can’t model human-in-the-loop (HITL) development here, what hope do we have for society?
The goal isn’t to hand over thinking to AI. It’s to work faster, think deeper, and stay human while using powerful tools. That’s the tension I want to explore — and the reason this site exists.
That’s also why I say: Trust your GUT — Generative Untrained Transformer.
It’s not a model. It’s a mindset. A reminder that insight doesn’t always come from training data. Sometimes it comes from friction, uncertainty, or a moment that just feels true.
Want to chat? Email me at richard@untraineddata.com