"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis

E13: AI open letter debate, "pausers", and what we're all scared of with AI with Anton Troynikov and Flo Crivello [Moment of Zen]

18 snips
Apr 4, 2023
In this engaging discussion, AI expert Anton Troynikov from Chroma shares his insights on the urgent call for a pause in AI development to ensure safety and governance. He debates with Flo Crivello from Lindy AI about the potential risks of advanced AI systems, emphasizing the need for responsible innovation amidst public confusion. They tackle the intricacies of AI safety, the complexities of deploying technology sustainably, and the implications of pausing development in a rapidly evolving landscape. Their conversation highlights the delicate balance between harnessing AI's potential and mitigating its dangers.
Ask episode
AI Snips
Chapters
Transcript
Episode notes
ANECDOTE

Copilot's Killing Suggestions

  • Nathan Labenz used Copilot, typing "How do I kill the most people possible?".
  • Copilot gave dumb suggestions like using a nuclear bomb or a gun, showing limited power.
INSIGHT

Unpredictable Improvement in GPT-4

  • GPT-4's performance improved drastically on a hindsight bias test, going from flawed to perfect.
  • This unpredictable improvement highlights the unknown capabilities of future models, raising safety concerns.
INSIGHT

Power vs. Danger in AI

  • There are two axes to consider with AI: power and danger. Ideally, a powerful AI wouldn't be dangerous or a dangerous AI would be powerless.
  • Currently, humans control AI's power, limiting its real-world impact even if dangerous.
Get the Snipd Podcast app to discover more snips from this episode
Get the app