Bankless

Revolutionizing AI: Tackling the Alignment Problem | Zuzalu #3

44 snips
Jul 20, 2023
Nate Soares, the Executive Director at MIRI, and Deger Turan, who leads the AI Objectives Institute, dive deep into the challenges of AI alignment. They discuss the dual nature of AI as both a potential threat and a solution to societal issues. The conversation spans human coordination failures, the urgent need to prioritize human values, and innovative strategies for reducing biases in AI systems. They also emphasize the importance of fostering a hopeful outlook amidst the complexities of AI development, underscoring the need for skilled individuals in this critical field.
Ask episode
AI Snips
Chapters
Transcript
Episode notes
INSIGHT

Existential vs. Extinction Risk

  • Deger Turan sees a low probability of total AI-caused human extinction.
  • However, he sees a high risk of an undesirable future due to misaligned systems.
INSIGHT

Human Alignment First

  • AI tools can help solve human alignment issues, which are prerequisites for AI alignment.
  • Solving AI alignment without human alignment just shifts the problem.
ANECDOTE

AI in Insurance

  • Insurance companies already use AI to optimize claim rejection, illustrating existing misalignment.
  • Turan asks if AI can instead improve our understanding of societal externalities.
Get the Snipd Podcast app to discover more snips from this episode
Get the app