Foundation Models for Scientific Discovery: From Paradigm Enhancement to Paradigm Transition
Fan Liu, Jindong Han, Tengfei Lyu, Weijia Zhang, Zhe-Rui Yang, Lu Dai, Cancheng Liu, Hao Liu
2025-10-20
Summary
This paper explores how powerful AI models, called foundation models like GPT-4, are changing how scientific research is done, going beyond just making things faster.
What's the problem?
Traditionally, science relies on researchers forming hypotheses, designing experiments, and interpreting results. The question this paper tackles is whether these new AI models are simply tools to *help* with these existing steps, or if they're actually causing a fundamental shift in *how* science itself is approached and conducted.
What's the solution?
The authors propose a three-stage framework to understand this shift. First, AI enhances existing scientific workflows. Second, AI becomes a true partner, helping scientists brainstorm problems and find solutions. Finally, AI could potentially make discoveries independently, with minimal human help. They then look at current examples of AI in science to see how these stages are playing out and discuss potential risks and future possibilities.
Why it matters?
Understanding how AI is changing science is crucial for researchers. This paper provides a way to think about this transformation and encourages the scientific community to consider what the future of discovery might look like with these powerful new tools.
Abstract
Foundation models (FMs), such as GPT-4 and AlphaFold, are reshaping the landscape of scientific research. Beyond accelerating tasks such as hypothesis generation, experimental design, and result interpretation, they prompt a more fundamental question: Are FMs merely enhancing existing scientific methodologies, or are they redefining the way science is conducted? In this paper, we argue that FMs are catalyzing a transition toward a new scientific paradigm. We introduce a three-stage framework to describe this evolution: (1) Meta-Scientific Integration, where FMs enhance workflows within traditional paradigms; (2) Hybrid Human-AI Co-Creation, where FMs become active collaborators in problem formulation, reasoning, and discovery; and (3) Autonomous Scientific Discovery, where FMs operate as independent agents capable of generating new scientific knowledge with minimal human intervention. Through this lens, we review current applications and emerging capabilities of FMs across existing scientific paradigms. We further identify risks and future directions for FM-enabled scientific discovery. This position paper aims to support the scientific community in understanding the transformative role of FMs and to foster reflection on the future of scientific discovery. Our project is available at https://github.com/usail-hkust/Awesome-Foundation-Models-for-Scientific-Discovery.