Knowledge Engineering Outperforms Data-Driven Learning in Specific AI Tasks

Data-driven learning is key in modern AI, but knowledge engineering, which formally encodes concepts using rules, can be superior in certain cases. According to V. Cheng & Z. Yu, people outperform chatbots in basic arithmetic because they use rules, not just examples. Knowledge engineering excels where rules are available, accuracy is vital (like in autonomous systems), and clarity is crucial (such as in education). Formalized knowledge, unlike evolving natural language, remains stable across cultures and languages, essential for both machine and human knowledge preservation. However, knowledge engineering is often overlooked in AI research, with few datasets formally encoding human knowledge for machine learning. AI should incorporate concept formation to develop more human-like reasoning.