Knowledge curation (not search) is the AI big data problem (www.daft.ai)

The article highlights a pressing challenge in the AI/ML community: the need for effective knowledge curation in enterprise and personal data. Unlike the public web, where information is structured through links and citations, private data often lacks context, hampering AI agents' ability to synthesize meaningful insights. This gap can lead to poor performance, as AI models operate like junior employees fumbling with fragmented information without the necessary background to make contextually aware decisions. The piece argues that while tools for raw data retrieval exist, the real hurdle is creating a knowledge layer that organizes and updates contextual information, enabling AI systems to derive accurate insights. Significantly, advancements from companies like OpenAI and Anthropic, such as the new "Company Knowledge" feature in ChatGPT, are paving the way for this transformation, allowing AI to access and dynamically sync with private data sources. The authors suggest that what lies ahead is a renaissance of "Expert Systems," where AI combines a robust knowledge base with inference capabilities drawn from improved language models. This shift aims to automate the synthesis of unstructured data and drastically elevate the effectiveness of AI applications, making knowledge curation a key differentiator for success in the AI landscape.