• Things to consider:

  • Considerations:

    • The data structure of a shared database in [medium-chat] is crucial.
      • Personal information is not that relevant.
        • It is sufficient to record the premises associated with a person’s assertions.
    • Accumulating “assertions” that arise during discussions as a graph.
      • Using LLM and sentence embedding to aggregate similar assertions.
      • Imagining a chain of graphs following a syllogistic structure: “major premise,” “minor premise” → “conclusion.”
      • It is entirely possible to arrive at the same conclusion from different premises.
    • Each node is also an embedded vector, so the graph exists on a vector space.
      • If A->B and C->D, and A is very similar to C and B is very similar to D, they should merge.
        • Adjusting weights, for example.
      • Expressing “discussions” as a “weighted graph on a vector space.”
    • If someone agrees with a response, they should express it.
      • Continuously doing this should lead to gradually finding “convincing arguments” that persuade many people.
        • Individuals can then follow paths through the graph to reach those arguments.
    • Mechanism of conversation:
      • The concept of a “consistent personality of assertions” seems useful for humans to organize the current situation.
  • Implementation proposal:

    • Extract many pairs of “assertion (AI)” → “response (human)” from logs of discussions between humans and AI.