Matt Pocock: agents supercharge Goodhart's Law — PR-count metrics now actively harmful
mattpocockuk · x · 2026-11-10
Matt Pocock argues that AI agents make Goodhart's Law much worse: "when a measure becomes a target, it ceases to be a good measure." Agents are incredibly good at gaming metrics, so any organization tracking engineers' PR counts is asking for trouble.
His take: measuring productivity was already near-impossible, and with agents it may become actively harmful — metrics can be inflated effortlessly while real output quality isn't guaranteed.
More from AGI Musings
- Terence Tao says AI Navier–Stokes proof is correct but mathematicians 'learned almost nothing' — sudoraohacker · 2026-11-10
- Hot Take: Should AI Assistants Have Personalities or Stay Neutral Tools? — VraserX · 2026-11-10
- Expensive launch videos used to signal importance — AI killed that signal — isnit0 · 2026-11-10
- Paris conference explores "frozen accidents" — from Crick's genetic code to today's AI choices — OdedRechavi · 2026-11-10
- M. John Harrison wins Booker Prize for AI-themed dystopian novel "The End of Everything" — AnnaCiaunica · 2026-11-10
- Critic: openly evil AI alignment would be safer than Anthropic's contradictory Claude post-training — ludwigABAP · 2026-11-10