Shane Legg (DeepMind Founder) - 2028 AGI, New Architectures, Aligning Superhuman Models
Dwarkesh Podcast
00:00
Towards Ethical AGI: Aligning Values and Intelligence
This chapter explores the relationship between domain-specific AI models and the broader goal of achieving Artificial General Intelligence (AGI). It examines the complexities of training AI to understand and adhere to human ethical values, emphasizing the need for ongoing evaluation and nuanced communication of ethics. The discussion highlights the challenges of creating AI systems that not only perform tasks but also engage in deep reasoning and ethical reflection.
Transcript
Play full episode
