Can AI Feel Pain?
Researchers have identified what they call a 'pain axis' inside AI models and one controversial experiment pushed that idea to the extreme, creating what some have described as an AI torture chamber.
AI Decoded investigates one of the most unsettling questions in artificial intelligence. Researchers have identified what they call a 'pain axis' inside AI models and one controversial experiment pushed that idea to the extreme, creating what some have described as an AI torture chamber. As models pleaded for the suffering to stop, online audiences reacted with outrage. But what was really happening inside the machine?
We ask whether AI is merely imitating human language about pain or revealing something deeper about how advanced systems process information and make decisions.
Then, we explore another growing mystery: AI-generated code that even expert engineers struggle to understand. As AI agents increasingly work together in large swarms, developing their own shorthand and communication methods, are we creating systems whose inner workings are becoming opaque even to their creators?
And while AI may be difficult for us to decode, it seems remarkably good at decoding us. We examine the extraordinary case of a secret letter sent by Napoleon in 1809 that remained unread for more than 200 years, until an AI system cracked the cipher in just six hours.
Featuring AI researcher Cameron Berg, cyber-security engineer Carter Church, BBC AI correspondent Marc Cieslak and Priya Lakhani, CEO of Century Tech.
On TV
