Abstract
It is likely that AI systems will eventually (or perhaps soon) be able to behave just like a conscious being, and in this paper I argue that we have sufficient moral reason to avoid such a future. Since the only evidence that we have for a thing being conscious is its behavior, we should grant corresponding moral status to beings that behave in a conscious manner, such as advanced AI systems. As soon as these “conscious” AI (CAI) exist, an ethical dilemma will arise that may not be easily addressed. On the one hand, conscious AI will likely consume a significant amount of energy such that they will be environmentally unsustainable. Current AI systems already consume a significant amount of natural resources, and it’s reasonable to believe that an AI with more general capabilities—like a CAI—would demand more. In order to prevent an environmental catastrophe, it may become necessary to “kill” CAI by shutting them down. On the other hand, as conscious beings it would be morally wrong to end their existence in this way, since it would be an instance of treating them as a means. It may also be instrumentally wrong because killing them would harm humans who form a valuable emotional bond with them. The existence of CAI would likely force us to choose between saving our own species or wiping out a new one of our own creation. It’s better to just keep Pandora’s box closed.
Details
Presentation Type
Paper Presentation in a Themed Session
Theme
Economic, Social, and Cultural Context
KEYWORDS
AI, CONSCIOUSNESS, CLIMATE CHANGE, ENVIRONMENT, ETHICAL BEHAVIOURISM
