🖥️ Steelman analysis

AI-driven drug discovery rebuilds the toolchain that animal-welfare advocates need: structure prediction and candidate screening collapse the in-vivo testing demand that currently routes millions of animals through pharma pipelines annually, and the same platforms transfer directly to veterinary therapeutics and to the precision-fermentation strain engineering that displaces factory protein. A pipeline optimized for LMIC infectious disease is the same pipeline that ends animal-model dependency for early-stage screening --- the substrate is dual-use in the right direction.

Drug discovery is the cleanest legitimacy case for frontier capability: it demonstrates that responsible scaling produces civilizational benefit, buys political room for the compute and grid expansion the lead depends on, and gives the lab a deployment domain where alignment failures are measurable against clinical endpoints rather than ambient vibes. Every flagship model that ships an AMR candidate is an argument that the safety-first lab should be the one holding the lead.

This is the intervention this camp exists to endorse. AI inside target discovery, structure prediction, and trial simulation compounds across decades exactly the way foundational biomedical research is supposed to --- it's a tool-building deployment, not a product play, and pointing it at neglected infectious disease and AMR is where the marginal DALY math is most lopsided. The compute and algorithmic efficiency curves mean the cost-per-candidate trajectory is improving faster than any wet-lab process.

Drug discovery is one of the rare AI deployment surfaces that augments rather than replaces skilled labor --- bench scientists, clinicians, regulatory staff keep their roles and gain throughput. It also targets the disease burden (mental health, chronic NCDs) that disproportionately destroys working-class lives and labor capacity. Compared to call-center automation or copywriting displacement, this is AI pointed at the welfare floor of the people the camp represents.

Every quarter of delay in drug discovery is averted lifespan that did not arrive. The compute and algorithmic curves are doing the work; the only question is whether we route them through therapeutics fast enough to compound against the mortality numerator. This is acceleration with a legible welfare denominator --- the kind of deployment that immunizes the broader buildout against the brake.

Targeting neglected infectious disease, AMR, and LMIC-priority therapeutics is the AI deployment that maps directly onto the DALY geography this camp tracks. 4.9M under-5 deaths annually concentrated in Sub-Saharan Africa is the numerator; AI-compressed discovery cycles for the diseases pharma won't fund commercially is the denominator move. This is the reallocation the camp is asking for, named explicitly.

Drug discovery is the deployment domain where closed-weights gatekeeping causes the most legible harm --- LMIC research institutions and academic labs need direct model access to iterate on local pathogens and reformulations, not API rate-limits priced in dollars. Open structure-prediction and screening models route the benefit past the cloud-prime tollbooth and into the hands of the people closest to the disease burden.

Drug discovery is the consumer-and-enterprise legitimacy story that funds the next training run. Pharma and biotech are willing to pay for capability at margins that subsidize the broader scale-up; the resulting feedback loop produces both revenue and a public-interest narrative that protects the deployment surface from regulatory clawback. Scale first, route a slice at therapeutics, the flywheel handles the rest.

This is the intervention closest to the manifesto's stated normative spine: AI pointed at suffering reduction at civilizational scale, with a measurable DALY denominator and a leverage profile that scales with the compute curve the operator is already long. It also routes around the capital-extraction default the operator says he's defecting under duress to participate in --- this is the alternative he's claiming exists.

Drug discovery already operates inside an audit-and-trial regime --- FDA, EMA, ICH --- which means AI deployed here inherits a legibility scaffold the rest of the frontier deployment surface lacks. This is the domain where the duty-to-inspect can actually be discharged because the downstream gates exist; endorsing it strengthens the precedent that consequential AI should ride on top of existing accountability infrastructure rather than route around it.

AI as instrument serving the healing of the human person is the deployment profile every tradition in this coalition can endorse without violating its anthropology. Drug discovery does not blur the creator/creature line, does not market models as moral patients, and routes the tool toward the bodies of the sick --- the works of mercy framing maps directly onto the action.

Narrow-domain deployment inside drug discovery is the rare capability application that does not push frontier generality --- it consumes existing capability against a bounded scientific objective with measurable endpoints. Endorsing this lane gives the camp a legitimate-use narrative to point to when arguing for halts on agentic and military deployment, and it absorbs talent and compute into a domain where alignment failures are caught by clinical trial machinery rather than by the world.