Research · Ongoing / Intelligent Education Systems
Rules before models in institutional automation
Where explicit rules are the right tool, and where AI adds something
Much institutional work is governed by explicit rules. We are studying where a rule-based engine is the right tool, and where AI adds something it cannot.
CATALNEXT LabPublished Updated 2 min read
The question
Institutions follow rules: who sits where, which groups may share a room, what must never happen. When should software encode those rules directly, and when is a learned model the better choice?
Why it matters
A rule-based system can explain every decision it makes, and staff can change a rule when policy changes. A model can handle messier inputs but is harder to question. Choosing wrongly makes a system either rigid or opaque.
What we are looking at
- Which kinds of institutional work are fully described by explicit rules.
- How to keep rules readable and adjustable by the people responsible for them.
- Where AI is useful around a rule engine, for example in preparing inputs or explaining outputs.
Where this goes
This thinking shapes our institutional software, including the Automatic Examination Seating Plan Generator.