Biodiversity as Information

In the AI Story

The framing has taken on new weight in the AI era. Every advance in AI's ability to process, recombine, and generate insight from data increases the marginal value of data that cannot be produced by any human or computational process. AI generates solutions by recombining existing information from the human corpus — the corpus of human-produced data that represents a tiny, recent, heavily biased sample of the information the biosphere contains. The biosphere's information was not produced by human observation. It was produced by evolution, running continuously for four billion years on every organism in every environment across every ecological challenge the planet has presented.

AI cannot access this information unless human beings first extract it — through genomic sequencing, biochemical analysis, ecological observation, and the slow, patient, funding-dependent work of field biology that Wilson championed and that the modern research university has been defunding for decades. Every species that goes extinct before it is studied takes its information permanently out of reach. The library closes. The volume burns. The information it contained — including the solutions to problems no human has yet learned to formulate — is gone.

The pharmaceutical example makes the stakes tangible. Natural products derived from organisms have produced roughly half of approved pharmaceuticals. The rosy periwinkle's vincristine. The Pacific yew's taxol. The cone snail's ziconotide. None was designed. All were discovered — extracted from organisms whose evolutionary history had produced the specific molecular configurations. AI-accelerated drug discovery operates on the space of known chemistry, the compounds already identified and entered into databases. The space of unknown chemistry — compounds produced by organisms not yet studied, or that will go extinct before they can be — is orders of magnitude larger, and it is shrinking every day.

The consilient insight is that the AI transition does not reduce the value of biodiversity; it dramatically increases it. The more powerful the tools for processing information become, the more valuable the information substrate becomes, and the more catastrophic its loss. This connects conservation to AI directly — not as metaphor but as infrastructure. Biological diversity is the raw material from which future pharmaceutical discoveries, biochemical innovations, and evolutionary insights will be extracted. Losing it before extraction is losing forever.

Origin

Wilson developed the informational argument across his late work, including The Diversity of Life (1992) and The Future of Life (2002), culminating in the Half-Earth synthesis (2016). The framing has been extended by researchers including Paul Ehrlich and Thomas Lovejoy, and by bioprospecting advocates who emphasize the pharmaceutical dimensions.

Key Ideas

Each species is a library. The metaphor is structurally precise: every organism encodes evolutionary solutions in configurations that cannot be reconstructed once destroyed.

Evolution is computation. The process that produced biological information operated across timescales and parameter spaces that no human computational process can replicate.

AI raises the stakes. The value of irreplaceable biological information rises as the tools for processing information grow more powerful.

Extraction is the bottleneck. AI can access biological information only after human research has extracted it. The rate of extinction now exceeds the rate of extraction.