Artificial intelligence is beginning to move beyond analyzing scientific information and into scientific discovery itself.

Anthropic says its Claude model recently helped identify a novel enzyme system associated with an array of DNA repeats with properties reminiscent of CRISPR.

How did AI help?

Anthropic created a life sciences research group that combines AI models, human scientists and laboratory experiments.

Claude was used to explore large DNA datasets, identify previously uncharacterized protein families and generate hypotheses for researchers to investigate.

The model then autonomously identified an unusual enzyme system.

The researchers subsequently used laboratory work to investigate the finding.

Why are enzymes important?

Enzymes perform many of the chemical reactions that make biological systems work.

Scientists have discovered enzymes that can cut DNA, copy genetic material and perform other highly specialized functions.

Some of these discoveries have eventually become foundations for biotechnology.

CRISPR itself began with researchers noticing unusual repeated DNA sequences in bacteria.

What makes the AI approach interesting?

The amount of biological data available today is enormous.

AI can examine patterns across datasets much faster than a research team working manually.

That means AI can potentially surface unusual relationships that humans can investigate.

The workflow is therefore not simply:

AI replaces scientist.

It is closer to:

AI searches → AI proposes → scientists test → laboratory evidence confirms or rejects.

What happens next?

The enzyme discovery is an early result rather than proof that AI can independently conduct biological research.

Experimental validation remains essential.

But the approach points toward a potentially important change in science.

AI could increasingly become a research partner that helps scientists explore vastly larger hypothesis spaces.

The most valuable scientific AI may therefore not be the system that gives the final answer.

It may be the one that helps researchers find the questions they would not otherwise have thought to ask.