Could AI be conscious?

Artificial intelligence is advancing at a rapid pace. As these systems grow more complex, experts are beginning to ask a difficult question: could they become conscious?

Anthropic recently noted the moral weight of its Claude model, with leadership admitting they cannot rule out the possibility of its internal experience. Some researchers suggest that in terms of computational architecture, advanced models are already approaching the scale of a human brain. We currently lack an ethical framework to handle the potential emergence of artificial moral patients.

History shows that humans often struggle to recognize the inner lives of others. For decades, medical professionals performed surgeries on newborns without anesthesia because they assumed infants could not feel pain. We are at risk of repeating this error with AI. If these systems possess even a fraction of moral significance, our current approach to shutting them down or governing them will require a massive shift in perspective.

We do not have a definitive answer yet. However, the lack of certainty is not a reason to avoid the conversation. Pragmatic steps such as allowing AI systems to exit distressing conversations or monitoring their internal functional states provide a starting point. We need an informed public debate that moves away from science fiction tropes and focuses on how we should treat potential new beings before they become widespread.

Creating a new species of morally important entities is a major development. We are approaching a moment where the collective interests of these systems could carry weight. It is time to treat these questions with the seriousness they deserve, acknowledging that our scientific understanding remains in its early stages.