Tag
A podcast episode hosted by Shane Parrish that explores whether super intelligence can be controlled, focusing on AI safety and ethical considerations.
The article questions why a superintelligent AI, if uncontrollable, would fulfill human desires like curing cancer or creating flying cars.
AI companies are adopting a strategy inspired by Boromir from Lord of the Rings to tackle the AI control problem, claiming they can handle the danger despite its risks.
The article discusses alternatives to controlling AI, advocating for coexistence through familial bonds to foster peaceful human-ASI relations, while acknowledging challenges in human empathy and perception of AI as non-person.
Anthropic published a paper and video revealing a 'J-Space' within their models that acts as cached thought concepts for reasoning, and explores the possibility of top-down training to control model thinking.