Thomas Kwa comments on Preventing an AI-related catastrophe—Problem profile