Continuing to address the challenges of AI safety, Rob Miles discusses a paper from the Machine Intelligence Research Institute (MIRI).
