Biologist Rebuts Claims That AI Could Enable Non-Experts to Create Supervirus

A sharp debate has emerged over whether artificial intelligence could allow an untrained individual to engineer a global biological catastrophe. The clash pits the computational predictions of economists against the physical realities of synthetic biology, highlighting a fundamental divide in how existential risks are calculated.

Writing on the 'Genes, Minds, Machines' blog, biologist Claus Wilke published a detailed rebuttal titled "I'm sorry, you're not going to die from an AI-engineered supervirus" to economist Noah Smith. Smith had previously posited in his article, "Here’s how we’re all going to die," that by 2029, AI could lower the barrier to entry for creating lethal pathogens to such an extent that a single disgruntled teenager could design and deploy a supervirus capable of killing 90% of humanity.

The Wet Lab Bottleneck

Wilke contends that Smith's scenario is scientifically implausible because it treats virus creation as a purely computational problem. According to Wilke, designing a genetic sequence on a computer is fundamentally different from successfully synthesizing a functional, stable, and transmissible virus in the real world.

He argues that biological synthesis requires a "wet lab"—a physical environment equipped with specialized hardware, high-quality biological materials, and a process of iterative laboratory testing. Wilke asserts that these physical requirements are constraints that AI cannot currently bypass, regardless of how advanced the software becomes. In his view, the necessity of physical experimentation remains the primary safeguard against the rapid deployment of engineered bioweapons.

The Expertise Gap

This disagreement reflects a broader tension between AI safety "doomers," who view AI as an immediate catalyst for biological apocalypse, and domain specialists who emphasize the friction of physical science. Wilke describes this disconnect as an example of "Gell-Mann amnesia," a phenomenon where an expert in one field finds the generalizations made by an expert in another field to be dangerously simplistic when applied to the former's area of expertise.

Regarding Smith's analysis of a topic close to his own professional background, Wilke stated, "I used to think Smith was a person with informed opinions, but reading his take on a topic close to my own expertise makes me question this premise."

Implications for AI Risk

The debate underscores the importance of domain-specific expertise when evaluating existential risks. If Wilke's assessment is correct, computational capabilities alone are insufficient to trigger a biological collapse, as the bottleneck is not the "blueprint" of the virus, but the physical act of building it.

While the scientific community continues to monitor AI's role in discovering new toxins or pathogens, the current divide suggests that the path from a digital sequence to a global pandemic is far more complex than a simple software prompt. The focus for policymakers may therefore need to shift from purely digital safeguards to the regulation of the physical materials and equipment required for synthesis.