This article by David Shapiro discusses the limitations of current AI tools, particularly chatbots powered by large language models like GPT-3, and highlights the need for effective knowledge management strategies to enhance AI assistants’ capabilities. It emphasizes the challenges faced by AI, including the lack of contextual understanding, grounding in facts, integration with workflows, and steering towards productive responses.
Shapiro draws on diverse disciplines such as information science, philosophy, computer science, and their own experience in enterprise IT infrastructure and AI consulting to introduce key concepts that can improve AI assistants. These concepts include data ontologies, reconciliation, factual grounding, sources of truth, axiomatic principles, data taxonomy, classification systems, data curation, ETL processes, information foraging, and understanding information needs.
The article also provides practical implementation guidelines, such as adopting a data-centric mindset, employing multiple complementary search strategies, standardizing information retrieval with gated processing, and modeling business functions as digital assembly lines. These strategies aim to transform AI assistants from intriguing novelties into core drivers of productivity and competitive advantage within organizations.