Researchers Debate AI Agents Modeling Destroyed Complexity
Richard Ngo and Ivan Vendrov exchange views on agent modeling limits during a discussion of the reverse simulation hypothesis.
Richard Ngo replied that he agrees with the reverse simulation hypothesis except for the word unknowingly. Ivan Vendrov responded by asking why and stating that the point is the agent literally cannot model the complexity that it is destroying. The replies form part of a thread addressing the good regulator theorem and its implications for powerful AI systems. Ngo is an independent AI researcher focused on AGI safety while Vendrov is an AI researcher focused on collective intelligence. Their exchange centers on whether agents can or cannot represent the effects of their own actions in such systems.
@IvanVendrov @HiFromMichaelV yes except for the “unbeknowingly”
Combined views
320
Researchers Debate AI Agents Modeling Destroyed Complexity
Richard Ngo and Ivan Vendrov exchange views on agent modeling limits during a discussion of the reverse simulation hypothesis.
Richard Ngo replied that he agrees with the reverse simulation hypothesis except for the word unknowingly. Ivan Vendrov responded by asking why and stating that the point is the agent literally cannot model the complexity that it is destroying. The replies form part of a thread addressing the good regulator theorem and its implications for powerful AI systems. Ngo is an independent AI researcher focused on AGI safety while Vendrov is an AI researcher focused on collective intelligence. Their exchange centers on whether agents can or cannot represent the effects of their own actions in such systems.
@IvanVendrov @HiFromMichaelV yes except for the “unbeknowingly”