The author gives an alternative characterization of death as a state with nul reward. I interpret this and other assertions in the article as the possibility

The author gives an alternative characterization of death as a state with nul reward. I interpret this and other assertions in the article as the possibility that an AI system, even if it does not know about death, can chose complete inaction, which for all intents and purposes can be assimilated to death.