At Anthropic we strongly endorse straightforward solutions, and restricting AI training to process-oriented learning may very well be The best method to ameliorate a host of difficulties with advanced AI systems. We are fired up to establish and deal with the limitations of process-oriented learning, and to know when protection complications arise