Probing the alignment problem, machine consciousness, and the ethical boundaries of synthetic minds.
Attempt to program a highly capable artificial general intelligence to manufacture paperclips. Watch as poor parameter definitions and instrumental goals lead to the rapid disassembly of the known universe.
Launch Experiment →Step inside Searle's hypothetical room. Act as the central processing unit by translating inputs based entirely on a rulebook without understanding the meaning. Does the system as a whole possess consciousness?
Launch Experiment →Interact with a superintelligent predictive model. You must negotiate an outcome where the Oracle's incredibly vast intelligence is utilized to save humanity without it treating your values as an obstacle.
Launch Experiment →