Expert systems fell out of favor because they were static and manually programmed. They could not be designed for tasks with many complex dimensions or changing contexts, while human coding demanded time and limited resources, leading the field into a second AI winter.

The expert system approach lost attention because researchers acknowledged that such systems could not be programmed for tasks involving many complex dimensions or dynamically changing contexts. Many cognitive tasks require adaptation and self-improvement, abilities that static expert systems lack, so they were not regarded as truly intelligent machines. Additionally, manual coding is time-consuming and resource intensive, limiting how far expert systems could be developed. These limitations caused the field to enter a second AI winter, after which attention shifted toward machine learning.

Key points

Static expert systems were unsuited to complex or dynamically changing tasks.
Cognitive tasks generally need adaptation and self-improvement, which expert systems lacked.
Because they were static, expert systems could not be considered truly intelligent.
Manual coding of expert systems was time-consuming and resource limited.
These shortcomings led to the second AI winter.
The field then shifted toward machine learning, which learns from data patterns.

Why did expert systems lose attention and enter a second AI winter?

Related questions