The strongest objection
More capable AI could accelerate discovery, improve treatment, reduce disability, and help solve urgent environmental and engineering problems. Delaying such progress could prolong avoidable suffering. People waiting for a treatment have a legitimate stake in decisions about the technology, and their interests should not be dismissed as a distraction.
Ask which capabilities are necessary
The relevant choice is not captured by comparing all technological progress with none. A prevention policy could preserve substantial scientific computing and bounded AI assistance. It should examine which benefits depend on prohibited general capabilities, which can be achieved with permitted tools, and how much delay an alternative route would create.
Some benefits may genuinely require capabilities that a strict ceiling would exclude. The prevention case must acknowledge that loss. It then has to explain why the reduction in catastrophic or irreversible risk justifies the cost, and how public investment could reduce that cost.
The calculation must also consider whether benefits reach people in practice. Discovery alone does not guarantee affordable treatment, equitable access, or lasting security. Those outcomes require institutions and resources alongside technical capability.
The hinge of disagreement
How large are the benefits uniquely dependent on the restricted capabilities, and what risk accompanies obtaining them? W360 argues that transformative benefits do not automatically justify an uncontrolled path to Artificial superintelligence: capability substantially beyond humans across a broad range of important tasks.. The strongest policy case would compare concrete research alternatives and uncertainty on both sides, rather than treating either the promised benefits or the feared harms as already settled.