Artificial intelligence (AI) is increasingly recognized as a transformative force in basic science, underscored by its prominent role in the 2024 Nobel Prizes in Chemistry and Physics.

Nobel laureates have emphasized that AI has the potential to significantly accelerate scientific discovery, positioning it as one of the most important technologies in history.

The appeal of AI lies in its ability to deliver results quickly and cost-effectively, as demonstrated by innovations like Sakana AI Labs' 'AI Scientist,' which claims to automate scientific discovery at a low cost.

While AI offers benefits such as faster and cheaper scientific results, it also raises concerns about public understanding, trust, and societal needs.

Although public trust in science remains relatively high, this trust is fragile and can be easily undermined by the complexities and contested nature of scientific evidence.

Engaging all stakeholders in discussions about the future of AI in science is essential to ensure that research aligns with societal needs and values.

In light of AI's rise, scientists are encouraged to reconsider the social contract between science and society, focusing on pressing issues in exchange for public funding.

Critical questions arise regarding the use of AI in science, including concerns about outsourcing integrity, environmental impacts, and the need to align research with societal expectations.

The integration of AI into science risks creating a monoculture of knowledge, prioritizing certain questions and methods over a diverse, transdisciplinary approach necessary for addressing complex social and environmental challenges.

Researchers face three key illusions when using AI: the 'illusion of explanatory depth,' where AI predictions do not equate to understanding; the 'illusion of exploratory breadth,' which limits the scope of hypotheses; and the 'illusion of objectivity,' which overlooks inherent biases in AI models.

Critics warn that the rapid production of research papers facilitated by AI could overwhelm the scientific ecosystem with low-quality content, straining the peer-review process.