NYTimes video asks why we cannot tell if an AI is conscious

In short: A New York Times video argues that sounding smart or emotional does not prove an AI has inner experience, and there is still no agreed test for AI consciousness.

What's going on

Carlos Lozada, in a New York Times Opinion video, looks at a question people often ask as AI gets better at talking, how would we know if an AI is conscious. He separates two ideas that are easy to mix up. Intelligence is being able to solve problems and use language well. Consciousness is having an inner point of view, meaning there is “something it feels like” to be you.

Lozada uses the 1959 novel A Canticle for Leibowitz to show how messy the topic can get. In the story, a skeptic wonders if he has a soul, and an abbot replies that he is a soul, not someone who merely has one. The point is that people can argue about words and definitions without agreeing on what evidence would settle the argument.

The video also points out a practical problem. Asking an AI “Are you conscious?” is not a reliable test, because a system can describe feelings without actually having any. It is like a puppet that can say “I am sad” without feeling sadness.

Some researchers suggest looking inside AI systems, not just at what they say. A 2023 report proposed a checklist based on scientific theories of consciousness, and found that the AI systems it reviewed were not strong candidates. The authors also said the checklist is not final and cannot prove the presence or absence of experience.

What to watch

As AI systems become more convincing, the pressure will grow to define what “consciousness” even means, and what kind of proof, if any, could separate real experience from good imitation.

Source: NYTimes