Hasty Briefsbeta

Bilingual

AI is impressive because we’ve failed at semantic web and personal computing

15 hours ago
  • Plain full-text search cannot answer complex queries like the one about a flag animal and historical dates, but ChatGPT can.
  • The success of LLMs is partly due to our failure to organize information with proper structure.
  • The pattern of abandoning structure and relying on search has become dominant in products like Google Drive, e-commerce, and documentation.
  • The Semantic Web and personal computing with structured semantic connections never fully materialized.
  • If knowledge were stored with rich semantic linking, simpler algorithms could answer queries with far fewer computational resources.
  • LLMs like ChatGPT are brute-force workarounds that infer structure from chaos, rather than elegant design.