knowledge graphs and llms

• good explanation on fundamental difference between knowledge graphs and llms

  • unifying large language models and knowledge graphs: a roadmap

  property	knowledge graphs	large language models
  representation	structured: [[triple	triples]]
  knowledge storage	[[explicit knowledge	explicit]]
  interpretability	high	low
  generalizability	low	high
  factual accuracy	high	low
  reasoning	symbolic and probabilistic	probabilistic
  knowledge update	easy	hard
  decisiveness	high	low
  unseen facts	no	yes
  incompleteness	high	low

• 

• create visualization

• insights are the following that knowledge graphs and llms are

  • fundamentally different
  • must interact in any intelligent system
  • key to superintelligence

• go on to dive into

  • cybergraph and truth machine
  • as beautiful counterpart probabilistic model for llm
  • in a modern soft3 stack