Expert Systems: AI Goes to Work

In the 1970s and 80s, AI found its first real customers by encoding expert knowledge as if-then rules.

25 min · Reviewed 2026

Knowledge You Could Write Down

By the mid-1970s, AI researchers pivoted from general intelligence to narrow expertise. The bet was that if you could interview a specialist, extract their decision rules, and encode them as if-then statements, the resulting program would match the expert on a well-defined task.

Edward Feigenbaum at Stanford led this shift. DENDRAL analyzed chemical mass spectra. MYCIN diagnosed bacterial infections and recommended antibiotics, reportedly outperforming some junior doctors in trials. XCON helped Digital Equipment Corporation configure VAX computers and saved the company tens of millions of dollars a year.

Anatomy of an expert system

A knowledge base of rules elicited from human experts
An inference engine that chains rules forward or backward
A working memory of current facts about the case
An explanation module that justifies conclusions in English

The field spawned an industry. Companies sold shells like KEE and ART, and knowledge engineer became a paid profession. Japan announced its Fifth Generation Computer Project in 1982, betting billions on parallel logic machines to dominate the coming AI era.

In the knowledge lies the power.
— Edward Feigenbaum

The big idea: symbolic AI genuinely delivered value for narrow, stable domains. Its collapse came not from failure but from the impossibility of scaling hand-written rules to everything humans know.

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-history-expert-systems-builders

What is the core idea behind "Expert Systems: AI Goes to Work"?
1. In the 1970s and 80s, AI found its first real customers by encoding expert knowledge as if-then rules.
2. Babbage designed a programmable machine using only mechanical parts
3. Natural language understanding
4. Claude Shannon turned communication into mathematics and gave AI the substrate i…
Which term best describes a foundational idea in "Expert Systems: AI Goes to Work"?
1. MYCIN
2. expert system
3. knowledge engineering
4. inference engine
A learner studying Expert Systems: AI Goes to Work would need to understand which concept?
1. expert system
2. knowledge engineering
3. MYCIN
4. inference engine
Which of these is directly relevant to Expert Systems: AI Goes to Work?
1. expert system
2. MYCIN
3. inference engine
4. knowledge engineering
Which of the following is a key point about Expert Systems: AI Goes to Work?
1. A knowledge base of rules elicited from human experts
2. An inference engine that chains rules forward or backward
3. A working memory of current facts about the case
4. An explanation module that justifies conclusions in English
Which of these does NOT belong in a discussion of Expert Systems: AI Goes to Work?
1. A knowledge base of rules elicited from human experts
2. An inference engine that chains rules forward or backward
3. Babbage designed a programmable machine using only mechanical parts
4. A working memory of current facts about the case
Which statement accurately describes an aspect of Expert Systems: AI Goes to Work?
1. Babbage designed a programmable machine using only mechanical parts
2. Natural language understanding
3. By the mid-1970s, AI researchers pivoted from general intelligence to narrow expertise.
4. Claude Shannon turned communication into mathematics and gave AI the substrate i…
What does working with Expert Systems: AI Goes to Work typically involve?
1. Babbage designed a programmable machine using only mechanical parts
2. Natural language understanding
3. Claude Shannon turned communication into mathematics and gave AI the substrate i…
4. Edward Feigenbaum at Stanford led this shift. DENDRAL analyzed chemical mass spectra.
Which of the following is true about Expert Systems: AI Goes to Work?
1. The field spawned an industry. Companies sold shells like KEE and ART, and knowledge engineer became a paid profession.
2. Babbage designed a programmable machine using only mechanical parts
3. Natural language understanding
4. Claude Shannon turned communication into mathematics and gave AI the substrate i…
Which best describes the scope of "Expert Systems: AI Goes to Work"?
1. It is unrelated to foundations workflows
2. It focuses on In the 1970s and 80s, AI found its first real customers by encoding expert knowledge as if-then rule
3. It applies only to the opposite beginner tier
4. It was deprecated in 2024 and no longer relevant
Which section heading best belongs in a lesson about Expert Systems: AI Goes to Work?
1. Babbage designed a programmable machine using only mechanical parts
2. Natural language understanding
3. Anatomy of an expert system
4. Claude Shannon turned communication into mathematics and gave AI the substrate i…
Which of the following is a concept covered in Expert Systems: AI Goes to Work?
1. MYCIN
2. knowledge engineering
3. inference engine
4. expert system
Which of the following is a concept covered in Expert Systems: AI Goes to Work?
1. MYCIN
2. expert system
3. knowledge engineering
4. inference engine
Which of the following is a concept covered in Expert Systems: AI Goes to Work?
1. expert system
2. knowledge engineering
3. MYCIN
4. inference engine
Which of the following is a concept covered in Expert Systems: AI Goes to Work?
1. expert system
2. MYCIN
3. inference engine
4. knowledge engineering

← Back to interactive lesson

Tendril · Builders · AI Foundations

Expert Systems: AI Goes to Work

In the 1970s and 80s, AI found its first real customers by encoding expert knowledge as if-then rules.

25 min · Reviewed 2026

Knowledge You Could Write Down

Anatomy of an expert system

A knowledge base of rules elicited from human experts
An inference engine that chains rules forward or backward
A working memory of current facts about the case
An explanation module that justifies conclusions in English

In the knowledge lies the power.
— Edward Feigenbaum

The big idea: symbolic AI genuinely delivered value for narrow, stable domains. Its collapse came not from failure but from the impossibility of scaling hand-written rules to everything humans know.

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-history-expert-systems-builders

What is the core idea behind "Expert Systems: AI Goes to Work"?
1. In the 1970s and 80s, AI found its first real customers by encoding expert knowledge as if-then rules.
2. Babbage designed a programmable machine using only mechanical parts
3. Natural language understanding
4. Claude Shannon turned communication into mathematics and gave AI the substrate i…
Which term best describes a foundational idea in "Expert Systems: AI Goes to Work"?
1. MYCIN
2. expert system
3. knowledge engineering
4. inference engine
A learner studying Expert Systems: AI Goes to Work would need to understand which concept?
1. expert system
2. knowledge engineering
3. MYCIN
4. inference engine
Which of these is directly relevant to Expert Systems: AI Goes to Work?
1. expert system
2. MYCIN
3. inference engine
4. knowledge engineering
Which of the following is a key point about Expert Systems: AI Goes to Work?
1. A knowledge base of rules elicited from human experts
2. An inference engine that chains rules forward or backward
3. A working memory of current facts about the case
4. An explanation module that justifies conclusions in English
Which of these does NOT belong in a discussion of Expert Systems: AI Goes to Work?
1. A knowledge base of rules elicited from human experts
2. An inference engine that chains rules forward or backward
3. Babbage designed a programmable machine using only mechanical parts
4. A working memory of current facts about the case
Which statement accurately describes an aspect of Expert Systems: AI Goes to Work?
1. Babbage designed a programmable machine using only mechanical parts
2. Natural language understanding
3. By the mid-1970s, AI researchers pivoted from general intelligence to narrow expertise.
4. Claude Shannon turned communication into mathematics and gave AI the substrate i…
What does working with Expert Systems: AI Goes to Work typically involve?
1. Babbage designed a programmable machine using only mechanical parts
2. Natural language understanding
3. Claude Shannon turned communication into mathematics and gave AI the substrate i…
4. Edward Feigenbaum at Stanford led this shift. DENDRAL analyzed chemical mass spectra.
Which of the following is true about Expert Systems: AI Goes to Work?
1. The field spawned an industry. Companies sold shells like KEE and ART, and knowledge engineer became a paid profession.
2. Babbage designed a programmable machine using only mechanical parts
3. Natural language understanding
4. Claude Shannon turned communication into mathematics and gave AI the substrate i…
Which best describes the scope of "Expert Systems: AI Goes to Work"?
1. It is unrelated to foundations workflows
2. It focuses on In the 1970s and 80s, AI found its first real customers by encoding expert knowledge as if-then rule
3. It applies only to the opposite beginner tier
4. It was deprecated in 2024 and no longer relevant
Which section heading best belongs in a lesson about Expert Systems: AI Goes to Work?
1. Babbage designed a programmable machine using only mechanical parts
2. Natural language understanding
3. Anatomy of an expert system
4. Claude Shannon turned communication into mathematics and gave AI the substrate i…
Which of the following is a concept covered in Expert Systems: AI Goes to Work?
1. MYCIN
2. knowledge engineering
3. inference engine
4. expert system
Which of the following is a concept covered in Expert Systems: AI Goes to Work?
1. MYCIN
2. expert system
3. knowledge engineering
4. inference engine
Which of the following is a concept covered in Expert Systems: AI Goes to Work?
1. expert system
2. knowledge engineering
3. MYCIN
4. inference engine
Which of the following is a concept covered in Expert Systems: AI Goes to Work?
1. expert system
2. MYCIN
3. inference engine
4. knowledge engineering

← Back to interactive lesson