Yes, a curated, domain-specialised contextual LLM is the anthesis of a general intelligence LLM, functioning as a narrow expert rather than a generalist.
Core Contrasts
- Scope: Narrow focus on a single field versus broad understanding across all human knowledge.
- Goal: Deep accuracy and strict rule-following in one area versus flexible reasoning and task-switching across many areas.
- Design: Small, gated data with high context versus vast, open-world data for general tasks.
Why They Clash
- General models try to know a little about everything.
- Specialised models ignore everything except one specific subject to be safe and correct.