Expert System or Artificial Intelligence
We are publishing a series of posts to do with ChatGPT and its use in the process safety discipline. The first posts in this series are:
These posts have generated some tricky questions. For example,
What is intelligence?
What is human intelligence?
What is artificial intelligence?
For a blog that discusses the narrow topic of process safety management we seem to have gotten into deep waters. Nevertheless, if we are to have confidence in the guidance that ChatGPT and related programs offer, we need to consider questions such as these. After all, most of what we read on the internet seems to have jumped straight to Question #3 without considering the first two questions.
Expert Systems
Although ChatGPT is referred to as Artificial Intelligence, it would probably be more accurate to refer to it as an Expert System. To gain guidance regarding this distinction, we asked the program the following question.
What is the difference between an expert system and artificial intelligence?
It replied as follows,
Artificial Intelligence (AI) is a broad field of computer science that aims to create machines or software that can perform tasks that typically require human intelligence. It encompasses a variety of techniques and approaches, including machine learning, natural language processing, computer vision, and more.
An expert system, on the other hand, is a specific type of AI system designed to emulate the decision-making ability of a human expert in a particular domain. Expert systems are built using knowledge-based techniques and typically rely on a knowledge base, which contains information and rules about a specific domain, and an inference engine, which uses this knowledge to make decisions or solve problems.
In summary, AI is a broader field that encompasses various approaches to creating intelligent machines, while an expert system is a specific application of AI that focuses on emulating human expertise in a particular domain using a knowledge base and inference engine.
I found that answer to be unsatisfactory; it does not tell us how to distinguish intelligence from expertise.
Heat Exchanger Design
Some years ago I worked with an engineer who developed an Expert System to do with the design of heat exchangers. The primary goal of the system was to check that a proposed design met the pertinent design codes. (Therefore, the first question the system asked was, ‘Where are you?’ Based on that reply it could identify the codes, standards and guidance to be followed.)
Not only did the expert system check for compliance, it would offer subjective critiques of the design. In other words, it could say,
Your design is code-compliant, but the velocity of the tube-side liquid seems to be too high.
Such a response is immensely valuable.
No one involved in this project ever thought of using the term Artificial Intelligence ― maybe because it was not applying intelligence, it was developing answers based on expertise.
I suggest that intelligence involves creativity and originality. Hence, in spite of what ChatGPT told me, an expert system such as the one developed by my colleague is not ‘a specific type of AI system’.
