Because it wouldn’t be a very intelligent move from the AGI. It’d be way easier for an AGI to set its own reward function to infinity by manipulating its own circuitry than it would be to warp the universe to its precise specifications. https://www.academia.edu/22359393/Utility_function_security_in_artificially_intelligent_agents
Because it wouldn’t be a very intelligent move from the AGI. It’d be way easier for an AGI to set its own reward function to infinity by manipulating its own circuitry than it would be to warp the universe to its precise specifications.
https://www.academia.edu/22359393/Utility_function_security_in_artificially_intelligent_agents