What kind of expert should a system be?

Area of Science:

• Artificial Intelligence
• Cognitive Science
• Human-Computer Interaction

Background:

• Expert systems rely on human knowledge, but experts struggle to articulate their reasoning processes accurately.
• This leads to 'reconstructed methods of reasoning' in AI, which may not align with user cognitive abilities.
• Authentic methods, derived from observing expert behavior, offer an alternative for AI development.

Purpose of the Study:

• To compare reconstructed methods of reasoning with authentic methods derived from expert behavior.
• To propose tests for verifying the authenticity of reasoning methods in expert systems.
• To illustrate the incorporation of authentic reasoning into expert systems using medical diagnosis examples.

Main Methods:

• Comparative analysis of reconstructed versus authentic reasoning.
• Development of criteria and tests to assess the authenticity of reasoning methods.
• Case study application in medical diagnosis expert systems.

Main Results:

• Reconstructed methods, while functional, often fail to mirror user cognitive processes.
• Authentic methods, based on observed expert behavior, show potential for improved user-system congruence.
• Proposed tests can differentiate between authentic and reconstructed reasoning.

Conclusions:

• Authentic reasoning methods are crucial for developing expert systems that are both proficient and user-centric.
• Integrating authentic methods enhances the usability and effectiveness of AI in complex domains like medicine.
• Further research should focus on refining methods for capturing and implementing authentic expert cognition.