AI is impressive because we've failed at personal computing
16 days ago
- #Semantic Web
- #AI
- #Information Organization
- Plain full-text search engines struggle with complex queries like identifying animals on flags linked to historical events.
- ChatGPT successfully answered a complex historical question about the flag of Dominica and British colonization in 1805.
- Google's AI widget failed to provide a correct answer to the same query.
- Modern LLM-based AI excels at extracting answers from unstructured internet data, compensating for poor information organization.
- Google Drive exemplifies the trend of abandoning structured organization in favor of full-text search, leading to a subpar user experience.
- The Semantic Web vision of structured, machine-readable data never materialized, leaving data unstructured and difficult for machines to interpret.
- The shift from structured HTML to dynamic JavaScript-driven content has further complicated machine readability.
- Personal computing has not evolved into personal knowledge bases with semantic connections, missing opportunities for structured knowledge management.
- Structured knowledge with rich semantic links would enable simpler, more efficient natural language processing without reliance on resource-intensive AI models.
- AI, particularly LLMs like ChatGPT, acts as a brute-force solution to infer structure from unstructured data, raising questions about the nature of knowledge.