20-year AI veteran Margaret Mitchell: don't slow the pace, change the game
mmitchell_ai · x · 2026-09-15
- Margaret Mitchell, an AI developer of 20 years, pushes back on the "slow the pace" framing: instead, we need to change the game.
- Key points:
- Safety and capability don't have to be separate; safety only slows things when tacked on outside the core technology.
- Instruction Tuning and RLHF—safety-informed alignment practices—partly created modern LLM chatbots, showing ethics can define rather than constrain technology.
- The "race" framing is awkward: no single track, no agreed finish line, no winner.
- She offers a garden metaphor: many things can grow and flourish at once, and well-done ethics can drive innovation instead of slowing it.
More from AGI Musings
- Andreessen: AI safety orgs are financially dependent on AI seeming dangerous — beffjezos · 2026-09-15
- zetalyrae: 'the next model will solve alignment' is a tiresome deus ex machina — zetalyrae · 2026-09-15
- AI safety should evolve from control paradigms into 'values engineering' — max_paperclips · 2026-09-15
- Nathan Lambert Predicts 'Robo-boss' as a Big Future Job Category for Humans — npew · 2026-09-15
- Closed APIs Are the New Latin, Open Weights the Vernacular Bible — alexcovo_eth · 2026-09-15
- 1557 Printing Monopoly Mirrors Today's Compute Thresholds — alexcovo_eth · 2026-09-15