The 'no plan' critique of AI alignment after The Curve

Writing after The Curve's third conference, an attendee calls the apparent AI safety strategy “No Plan”: align today's models, then have them help solve alignment for more capable systems. The writer argues mistakes in automated alignment could compound and says proposals to “pace the frontier” remain ill-defined. They favor limits on frontier-training compute to buy time, while doubting that better evaluations and monitoring alone would be enough.