I'm sorry, you're not going to die from an AI-engineered supervirus

TEXT START: Opinion writer and economist Noah Smith just dropped an article about AI-designed superviruses.

The Dissection

This is an expert rebuttal that shrinks AI bio-risk into a deliberately absurd package: an untrained teenager, in 2029, designs a virus that kills 90% of humanity, builds it, distributes it, and defeats every biological tradeoff. The author then uses genuine present-day constraints to imply that the entire category of AI-enabled biological catastrophe is similarly implausible.

The text is also attention arbitrage. It redirects readers from cinematic AI catastrophe toward neglected public-health risks, while converting domain expertise into authority and subscription-era polemic. It correctly attacks a cartoon. It then quietly treats that victory as a refutation of the underlying trend.

The Core Fallacy

The author mistakes current capability boundaries for permanent laws. Today's difficulty designing simple biological systems does not prove that future AI cannot automate the experimentation, search, synthesis, feedback, and specialist labor that currently make biology difficult. Under P1, cognitive bottlenecks are targets for automation, not sacred barriers.

The biological tradeoffs are real, but the article treats them as a veto rather than an optimization landscape. A threat need not be an impossible Ebola-meets-influenza supervirus or kill 90% of humanity to cause enormous systemic damage. The author attacks the most theatrical endpoint and then declares the broader risk defeated.

The same article condemns unsupported faith in AI capability while making its own unsupported prophecy: “virtually impossible,” “not for decades,” and “will never exist.” That is not a causal forecast. It is presentism wearing a lab coat.

Hidden Assumptions

• Current AI limitations will persist rather than compound through automated research loops.
• Biological design is a one-shot task rather than an iterative system involving experimentation, selection, and optimization.
• A rogue actor must be an untrained teenager rather than a state, wealthy organization, or technically supported network.
• Human behavior will respond quickly, rationally, and uniformly enough for isolation to contain severe outbreaks.
• Existing public-health institutions can coordinate globally despite legal, political, economic, and cultural inertia.
• “Not civilization-ending” means “not a major systemic threat.”
• Natural and engineered risks are competing distractions rather than overlapping preparedness problems.
• The author's credentials substitute for a demonstrated causal chain.

Social Function

Classification: partial truth, prestige signaling, and ideological anesthetic.

The text reinforces expert jurisdiction and familiar institutions: surveillance, vaccination, research, filtration, and public-health administration. Those measures are real and useful, but the article presents them as if institutional competence were a stable fact rather than a lagging defense. It reassures the audience that the machine has not yet breached the expert boundary, so the boundary can be treated as durable.

This is not pure copium. The criticism of the 2029 scenario, the difficulty of biological engineering, and the neglect of natural pathogens are materially valid. The anesthetic enters when temporary technical friction is promoted into a forecast of long-term safety.

The Verdict

The article is a competent demolition of a sensational story and a weak analysis of future capability. It correctly says that the supplied 90% scenario lacks a credible chain of design, construction, and distribution. It does not justify the leap from “not demonstrated today” to “virtually impossible for decades” or “never.”

Under the Discontinuity Thesis, AI does not need a supervirus to destroy the post-WWII order; P1–P3 already specify the direct route through cognitive automation and the collapse of productive participation. The article's deeper failure is therefore methodological: it mistakes the survival of today's biological bottlenecks and institutions for structural permanence. It is a polished rebuttal of a cartoon, not a refutation of discontinuity.