KM4BI

Ben Lamm on Why Biology Is Approaching a Singularity

Colossal’s cofounder explains how AI, comparative genomics, DNA synthesis, and ex utero development could make biology programmable.

Published

Biology is beginning to look less like a fixed inheritance and more like an engineering system. Ben Lamm, cofounder of Colossal, argues that advances in genome editing, artificial intelligence, sequencing, and reproductive technology are converging fast enough to change both conservation and the design of living organisms.

What AI can do now

Colossal initially used language models as connective infrastructure across lab notebooks, project management, and reporting. Lamm sees the next step in resolving the language problem inside science itself. Papers may describe the same gene differently, classify an experiment in incompatible ways, or become difficult to reproduce. AI can help reconcile those records, uncover overlooked links, and propose experiments.

It still cannot turn one genome into a reliable blueprint for an ideal organism. Lamm’s preferred route is comparative genomics at scale: combine thousands or millions of samples with known traits, evolutionary relationships, and environmental context. That is why he emphasizes global biobanks and describes curated biological data as potentially more valuable than the general-purpose model analyzing it.

From hundreds of edits to DNA synthesis

Lamm says Colossal routinely delivers more than 300 genetic edits at over 90 percent efficiency and is testing deliveries of 1,000 edits. He does not assume that curve will continue indefinitely. For larger changes, he expects direct DNA synthesis and large genetic cargo swaps to become more practical than ever-bigger multiplex editing runs.

His forecast is that within a decade it should be possible to specify important phenotypes in certain base organisms, create the required genome through editing or synthesis, and develop the animal outside the body. He also predicts that Colossal could produce a mammal gestated entirely ex utero within 24 months. These are company claims and timelines, not demonstrated endpoints, but they show where the technical race is heading.

De-extinction is also a governance problem

Choosing a species is not just a technical exercise. Colossal considers its ecological function, why it disappeared, its place in a food web, the views of Indigenous communities, and whether it can draw attention to conservation. The dire wolf project, Lamm says, was partly intended to connect popular culture with wolf conservation and genome engineering.

Animal welfare extends beyond birth. Social mammals need groups, learned behavior, and suitable habitat. Lamm describes support for elephant programs that study artificial herd formation, migration corridors, and social hierarchy — knowledge that could inform both present-day elephant conservation and any future mammoth population.

Backing up the biosphere

Lamm’s central appeal is to preserve existing life before debating every possible synthetic species. He warns of severe biodiversity loss over the next 25 years and argues that zoos and nonprofits cannot build the necessary infrastructure alone. Governments, researchers, and funders would need to create interoperable repositories for cells, genomes, and associated data.

In that framing, de-extinction is the visible edge of a broader project: learning how to store, interpret, and eventually restore biological diversity. The opportunity is enormous, but so is the need for public rules around who controls these tools and which uses society will accept.

Source