Why LLMs Can Seem So Dumb to Experienced Developers
Why we judge probabilistic systems using deterministic expectations, and why the model is often missing parameters we already carry in our heads.
Field notes
What I am learning while building.
I am documenting the engineering problems I keep running into with AI, what I tried, what failed, what changed and what I learned from building around those failures.
Series 01 · Learning to Engineer with AI
From probabilistic models to engineering systems
The opening series starts with the distinction that changed how I think about LLMs, then follows the engineering consequences one by one.
What comes next
Publish the lesson, not the private build plan.
Future field notes will stay unannounced until the underlying work is ready for public review.