People who know the most often sound the least certain
3 hours ago
- The conversation between AI experts tackled deep questions about AI automating research, setting its own objectives, and the limits of reward-based learning.
- Public reaction focused on verbal tics like 'like' rather than substance, revealing a preference for polished certainty over genuine reasoning.
- Truth in AI is complex and requires hedging, but public discourse rewards simple, confident statements, creating a dangerous gap between reality and perception.
- Formats like Dwarkesh Patel's podcast allow experts to reason in public, exposing the messy process of thinking and testing assumptions.
- The author offers practical tips for clear communication: simplify explanations, use pauses, tell stories, state claims before caveats, and write in public.
- Experts must balance accuracy with accessibility to influence public understanding, or risk being overshadowed by polished but shallow narratives.