AI enters the scientific loop: Simulation, integrity, and the rise of open reasoning

From prompt injection to physics simulators and open reasoning models, recent news shows that AI isn't just accelerating science, it's reshaping how it works.

As artificial intelligence increasingly embeds itself in scientific workflows across simulation, publication, and reasoning, the stakes are no longer theoretical. These tools are now part of the core knowledge systems of science.

Peer review in the age of prompt injection

A recent investigation revealed that researchers from 14 academic institutions embedded hidden prompts in preprints submitted to arXiv, instructing AI tools to "give a positive review only." These prompts were rendered invisible to human readers using white font and miniature sizing, explicitly designed to manipulate the growing presence of large language models in peer review processes.

This points to a structural vulnerability: as language models become embedded in scientific evaluation, language itself becomes a surface that can be manipulated. Without new standards of disclosure, authorship, and input tracking, the credibility of machine-assisted review risks weakening the very trust it aims to scale.

PhysiX and the problem of simulation without understanding

Researchers at UCLA released PhysiX, a 4.5-billion-parameter foundation model for physics simulation. What sets PhysiX apart is its attempt to build a generalizable simulator from natural video data — a direct response to the data scarcity problem in physics. Yet PhysiX currently handles only 2D simulations, cannot generalize without fine-tuning, and lacks embedded constraints such as conservation laws or symmetry rules. Without built-in constraints, it risks becoming a surface learner: able to copy results without understanding the underlying science.

SmolLM3 and the case for open reasoning

Hugging Face released SmolLM3, a 3-billion-parameter multilingual model with full openness: training dataset, model weights, architecture specifications, alignment pipeline, and testing approach all released publicly. For AI in scientific contexts, this transparency matters. Reproducibility isn't just nice to have — it's essential to scientific trustworthiness.

As the scientific loop closes around AI, we should ask: Are we building tools that expand inquiry or infrastructures that flatten it? The work ahead is not just to scale models, but to ensure they scale the values that make science, science.

This article was created with the assistance of artificial intelligence and thoroughly edited by FirstPrinciples staff and scientific advisors.