# What are the limits to biomedical research acceleration through general-purpose AI?

## TL;DR

This paper explores how much general-purpose AI can speed up biomedical research and finds that while some tasks could be accelerated significantly, practical limits and infrastructure challenges remain.

## Contribution

The study introduces a framework to evaluate GPAI's potential in biomedical research and identifies key bottlenecks to achieving strong acceleration.

## Key findings

- Current GPAI could provide a 2x speed increase, while future GPAI may enable up to 25x acceleration for physical tasks and 100x for cognitive tasks.
- Expert opinions suggest strong acceleration is unlikely for experiment design and execution but plausible for manuscript preparation and publication.
- Assimilation of GPAI tools by the scientific community is identified as a critical bottleneck to realizing GPAI's full potential.

## Abstract

Although general-purpose artificial intelligence (GPAI) is widely expected to accelerate scientific discovery, its practical limits in biomedicine remain unclear. We assess this potential by developing a framework of GPAI capabilities across the biomedical research lifecycle. Our scoping literature review indicates that current GPAI could deliver a speed increase of around 2x, whereas future GPAI could facilitate strong acceleration of up to 25x for physical tasks and 100x for cognitive tasks. However, achieving these gains may be severely limited by factors such as irreducible biological constraints, research infrastructure, data access, and the need for human oversight. Our expert elicitation with eight senior biomedical researchers revealed skepticism regarding the strong acceleration of tasks such as experiment design and execution. In contrast, strong acceleration of manuscript preparation, review and publication processes was deemed plausible. Notably, all experts identified the assimilation of new tools by the scientific community as a critical bottleneck. Realising the potential of GPAI will therefore require more than technological progress; it demands targeted investment in shared automation infrastructure and systemic reforms to research and publication practices.

The online version contains supplementary material available at 10.1038/s41598-025-32583-w.