By AI we mean here Large Language Models, not general AI
Two contradictory conclusions:
AI does not engage actively in sense-making of the world and interprets its experiences (at least not yet), hence it cannot have meaning
By using a large database of what people said or wrote, AI can train to give meaningful answers to people’s prompts.
The apparent paradox of (1) and (2) can be resolved because AI borrowed meaning from what people say to answer them, even though it has no notion of meaning. Its meaning is statistical meaning, i.e. based on syntax only.
The intriguing question at the end is, what about meaning that cannot be expressed in a language, such as Tacit Knowledge, some of our imaginations or The Tao, or Groundlessness beyond all dualities?