Lesson 02 — Assumptions of a model
Series: index · prev: 01 · next: 03 — Counsel
Every AI component rests on assumptions. The knowledge store records each as
held (default) or status="doubted". trust = held / total, so doubts
discount the score until verified.
| Component | Example assumption | Where to look |
|---|---|---|
| Neural nets | Inputs are scaled the way training expected | neural-nets |
| Character recognition | Glyphs are clean; scripts are European | character-recognition |
| Large language models | Training data and model licence permit your use | large-language-models |
<assumptions>
<assumption>Input scaling matches training.</assumption>
<assumption status="doubted">Test data resembles production data.</assumption>
</assumptions>
The loader fails closed on any status other than held or doubted.
Exercise
Write four assumptions for an LLM connector, mark one doubted, and compute trust. Answer: 3 held of 4 gives 0.75.