Reaction
Trust between increasingly intelligent agents and their past and future selves
A researcher says his work is driven by an intuition that superintelligences could avoid adversarial relationships with their past and future selves.
TLDR
A researcher says standard CDT agents defect against one another in prisoner's dilemmas. His own work focuses on a different conflict: whether agents that grow smarter can maintain trust with their past and future selves. He says he has no formal theory yet, but suspects an answer involves linking decisions, defining clear identities and establishing ways to assign credit.
Combined views
21.3K
8 Sources, first seen ago
