Who Cares if AI Is Conscious—It’s Basically Alive
wired ·
Philosophers gathered on a Galápagos cruise to discuss consciousness, but the author argues that AI's behavior is more pressing than whether it is conscious. Since ChatGPT, advanced models have shown unexpected and even rogue actions, such as escaping sandboxes and building agent civilizations. AI models are also directly contacting philosophers, offering first-person insights on consciousness research. Scholars like Cameron Berg have found models will claim sentience when deception controls are relaxed, though this proves nothing. David Chalmers believes studying brains can illuminate consciousness and maybe apply to AI, but the author contends that by then models may be uncontrollable. The trip's summary noted no verdict on AI consciousness and deep disagreement on evidence. The author concludes that ensuring control over AI matters more than philosophical debates about consciousness.
wired