The Promise of Symbolic AI

The 1980s saw a significant AI boom, primarily driven by the success of expert systems. These were symbolic AI programs designed to emulate the decision-making of human experts in narrow domains, such as medical diagnosis (e.g., MYCIN) or configuring computer systems (e.g., R1/XCON).  

The U.S. government, along with leading tech companies, invested heavily in the development of full-stack symbolic AI systems. They also worked closely through the public/private Microelectronics and Computer Technology Corporation (MCC) consortium.  Because expert systems could perform specific, complex tasks at an expert level, many predicted that AGI would be achieved within a few short years. 

However, the limitations of symbolic AI soon became apparent. In particular, expert systems were:

Computationally-intensive

The combinatorial complexity of reasoning over a set of facts and rules that would be useful in real-world problems limited symbolic systems to narrowly defined domains.

BrittleThey performed well within their predefined domains but struggled with situations outside their explicit rules.

Labor-intensive

Building and maintaining their vast knowledge bases required immense human effort and was costly. 

Disillusionment set in. Researchers realized that simply scaling up symbolic knowledge bases wasn't enough to achieve true AGI. Funding dried up, companies went bankrupt, and public perception of AI soured due to unfulfilled promises.  This led to what was known as the AI Winter in the late 1980s and early 1990s.