Quote · Modern Wisdom
The Collapse of American Politics - Ezra Klein - #1114
Where this was said
How Ezra Would Approach AI Safety
At 1:20:14 · chapter starts 1:16:34
Klein's concrete AI policy agenda begins with dramatically strengthening public evaluation capabilities — the capacity to audit, test, and understand AI systems — which Trump and Musk have gutted. Second and most personally, he would move quickly on children: he believes there is reasonable consensus on protecting children from AI companions and lovers, and his own experience as a lonely, bullied, smart nerdy kid makes the case visceral. If he had been able to disappear into frictionless digital relationships before developing real human connections, the friction that made him would have been removed. Third, a positive public-goods agenda for AI — directing its capabilities toward orphan diseases, IRS tax filing, government navigation — rather than only harm prevention. Fourth, requiring that AI systems always maintain a legible chain of reasoning in English, so humans can audit decisions before the 'black box gets too black boxy.' He also expresses concern about AI surveillance — both the macro panopticon and the micro: eye-tracking and productivity-monitoring software that turns workers into machines. Turning machines against people to make people more machine-like, he argues, is 'inimical to human flourishing.'
Klein was a lonely, bullied kid who fought through friction to find himself. He is genuinely scared about what would have happened if he'd had access to frictionless AI friends, tutors, and lovers during those years. That friction made him who he is. Removing it from childhood development is the AI intervention he thinks we most urgently need to regulate.
Klein recommends AI systems always keep a readable reasoning log in English so humans can audit their decision-making before the black box gets too black.
