AI Terminology Mapping
How common AI/ML terms map to AlignTrue concepts:
| Common AI Term | AlignTrue Equivalent | Learn More |
|---|---|---|
| Context graph | Trajectories + co-occurrence projections | Trajectories |
| Knowledge graph | Co-occurrence graphs, transition n-grams | Trajectories |
| RAG (Retrieval Augmented Generation) | Query artifacts (retrieval) + Derived artifacts (generation) | Query artifacts, Derived artifacts |
| Memory / Long-term memory | Projections | Projections |
| Reasoning trace / Chain of thought | Trajectories (hash-chained steps) | Trajectories |
| Audit trail / Receipts | Events + egress receipts | Envelopes |
| Embeddings | Deferred in v1 (structure-first approach) | Simulation |
| Provenance | Query/Derived artifact lineage | Derived artifacts |
Why different names?
AlignTrue prioritizes determinism and auditability. Traditional AI terms like “memory” or “context” are fuzzy. AlignTrue concepts are explicit:
- Trajectories are hash-chained, tamper-evident reasoning logs.
- Query artifacts prove what the AI saw (with hashes).
- Derived artifacts prove what it decided (with lineage).
Last updated on