Expert systems era

Definition

In simple words

Sit with a top doctor for months, and write down all their rules of thumb. Now anyone can ask the program. It works well, until a case comes up that the rules never covered.

Why it matters

It shows both the power and the limit of hand-written knowledge. Companies still use rule engines for tax, eligibility and compliance, because rules are exact and easy to audit.

How it works

Why they stalled

• Getting knowledge out of experts was slow and costly.
• Thousands of rules became impossible to maintain.
• They were brittle: a case outside the rules gave poor answers.
• They couldn't learn from new data by themselves.

In code

RULES = [
    ({"fever", "cough"}, "flu"),
    ({"flu", "body pain"}, "rest and see a doctor"),
]

def infer(facts):
    facts = set(facts)
    changed = True
    while changed:
        changed = False
        for needs, gives in RULES:
            if needs <= facts and gives not in facts:
                print("Rule fired:", needs, "->", gives)
                facts.add(gives)
                changed = True
    return facts

result = infer(["fever", "cough", "body pain"])
print("Advice:", "rest and see a doctor" in result)

Rule fired: {'cough', 'fever'} -> flu
Rule fired: {'body pain', 'flu'} -> rest and see a doctor
Advice: True

Common mistakes

• Writing off rules because they're old. They're right for clear, stable logic.
• Expecting a rule system to handle cases it has no rules for.

Check yourself

Summary

• Experts' knowledge becomes if-then rules.
• Clear and explainable, but brittle and costly.
• Rule engines are still used today.