Dubito Ergo Sum: Exploring AI Ethics

AI lacks the capacity for self-doubt, which the authors argue is fundamental to moral agency. Drawing on Descartes' philosophical framework, the paper contends that true morality requires the ability to question oneself—a capability current AI systems do not possess, suggesting AI cannot be genuinely morally responsible agents.

Human morality is primarily intuitive rather than rule-based. The paper notes that moral psychology research shows humans make moral decisions intuitively first, and only afterward use ethical theories to rationalize those decisions. This challenges the common approach of trying to program AI with explicit ethical rules.

Sensory experience and embodied understanding are crucial to human morality in ways AI cannot replicate. The authors emphasize that human moral knowledge is grounded in physical sensation and lived experience—dimensions that AI systems fundamentally lack, creating a significant gap between human and artificial moral reasoning.

Current AI ethics approaches may be fundamentally limited by not accounting for these deep differences between human and artificial cognition. Because AI cannot doubt itself, experience the world sensorily, or understand in the embodied way humans do, the paper suggests that traditional ethical frameworks may not translate directly to AI systems.

The paper deliberately leaves the problem unsolved rather than proposing a definitive answer. The authors acknowledge that after thousands of years of studying human ethics, we're only beginning to understand it—suggesting humility is warranted when attempting to implement ethics in AI systems.