Abstract: Itamar Arel
Reward Driven Learning and the Risk of an Adversarial
Artificial General Intelligence
Abstract: A myriad of evidence exists in support of the notion that learning in humans and other animals is driven by reinforcement learning (RL). Modern cognitive psychology as well as neuroscience findings strongly suggest that much of our behavior is driven by both positive and negative feedback received from the environments with which we interact. The notion of reward is not limited to physical indicators originating from the physical environment, but has been broadened to embrace signaling generated internally in the brain, based on intrinsic cognitive processes. Artificial General Intelligence (AGI), coarsely defined as human-level intelligence manifested over non-human platforms, is commonly perceived as one of the paths that may lead to the singularity. Such a path has the potential of being either constructive or devastating to the human race, greatly depending on what such AGI will be like. I will briefly discuss the perceived ramifications of an RL-based AGI reality.