AI may accelerate some scientific tasks dramatically, but Wellcome Sanger Institute director Matt Hurles argues that the gains could pile pressure onto the slower links around them: funding, regulation, peer review and coordination. His central point is less “AI will cure everything” than “what happens when one part of a research pipeline suddenly moves much faster than the rest?”
Watch Desk analysis
What happened
In a personal reflection published on the Wellcome Sanger Institute blog on 2 November, Hurles considers the prospect of AI making parts of science 10 to 100 times faster. Resultsense summarises his argument: scientific work is a chain of tasks, and AI’s capabilities are uneven. Speeding up analysis or idea generation will not automatically speed up experiments, access to biological materials, regulatory approval or decisions about funding.
Hurles points to signs of pressure already emerging, including growing volumes of written applications for jobs, PhD places and grants. He also describes a lack of detailed evidence about how research projects actually run. He says he could not readily say how much time the tasks in his own recent projects took, making it harder to know where AI might help or where queues might form.
Why it matters
If AI makes some stages faster, the queue may simply move downstream. More applications could strain review panels; faster research could meet slow approval processes; and automated experiments could create new demands for coordination and oversight. The practical question for research institutions is not only which tasks AI can accelerate, but whether the surrounding systems can absorb the extra work.
Hurles also warns that separate institutions may fix only the bottlenecks they can see or are equipped to address. Funders may focus on grant processes, while technology companies invest in tools and automated labs. That leaves a risk of important gaps between them.
Our read
This is a useful corrective to treating “science” as one giant button marked faster. Hurles makes a case for mapping how research really happens, including how long its tasks take and where they depend on one another. That is less glamorous than announcing a new model, but it may be what separates a genuine acceleration from a longer queue with a shinier sign.
What to watch
- Whether funders change grant processes as application volumes rise.
- Whether regulators adapt approval systems to new research methods and risks.
- Whether AI-linked lab automation produces evidence of faster experimental work.
- Whether researchers and institutions gather better data on how projects run from start to finish.
Discussion spark: If AI makes research tasks faster, should funders and regulators prioritise speeding up their own processes, or would that risk weakening review and oversight?
Sources and evidence
- Sanger director: AI-accelerated science could outrun funders, regulators and reviewers – Resultsense (18 November 2026, 09:10 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.