WORK
Escaping Brittleness
Holland's 1986 paper diagnosing the structural limitation of rule-based AI — systems whose hand-coded rules could not recombine to address novel situations — and proposing adaptive parallel rule systems that anticipated the deep learning revolution by thirty years.
Escaping Brittleness, published as a chapter in the 1986 collection Machine Learning: An Artificial Intelligence Approach, Volume II, made explicit what would become Holland's most consequential contribution to AI thinking. The paper argued that expert systems — the dominant AI paradigm of the 1970s and 1980s — were brittle because they did not discover building blocks. Their rules were hand-coded, fixed, incapable of recombination. They worked within their programmed domain and failed catastrophically outside it because they had no mechanism for generating novel combinations of existing knowledge. Holland proposed an alternative: parallel rule-based systems in which rules competed, combined, and evolved through variation and selection. The rules were building blocks. The system's intelligence emerged from their recombination. Thirty years later, large language models achieved what Holland was reaching for through a different technical mechanism but with the same structural logic.
In The You On AI Field Guide
The paper's diagnosis of rule-based AI anticipated its