The Hot Mess of AI (Mis-)Alignment

The Hot Mess of AI (Mis-)Alignment

From Linear Digressions by Katie Malone

March 23, 2026 · 23 min

About this episode

This episode discusses new research on AI misalignment, suggesting it may be less about evil intentions and more about distraction and complexity.

The paperclip maximizer — the classic AI doom scenario where a hyper-competent machine single-mindedly converts the universe into office supplies — might not be the AI risk we should actually lose sleep over. New research from Anthropic's AI safety division suggests misaligned AI looks less like an evil genius and more like a distracted wanderer who gets sidetracked reading French poetry instead of, say, managing a nuclear power plant. This week we dig into a fascinating paper reframing AI misalignment through the lens of bias-variance decomposition, and why longer reasoning chains might actually make things worse, not better. - "The Hot Mess Theory of AI Misalignment: How Misalignment Scales with Model Intelligence and Task Complexity" — Anthropic AI Safety. https://arxiv.org/abs/2503.08941

People in this episode

Host: Katie Malone

Topics covered

  • AI alignment
  • machine learning
  • bias-variance decomposition
  • AI safety
  • task complexity

Keywords

  • AI misalignment
  • paperclip maximizer
  • Anthropic
  • bias-variance
  • reasoning chains

Mentioned in this episode

Organizations: Anthropic

Books & works: The Hot Mess Theory of AI Misalignment: How Misalignment Scales with Model Intelligence and Task Complexity

More episodes of Linear Digressions

Explore listener stats, chart rankings, contacts and more on the Linear Digressions podcast page.