In line with the general trend incorporates artificial intelligence into almost all areasresearchers and politicians are increasingly using TO models trained on scientific data to infer answers to scientific questions. But can artificial intelligence eventually replace scientists?
On November 24, 2025, the Trump administration signed an executive order announcing Genesis missioninitiative to build and train a range AI agents on federal scientific data sets “to test new hypotheses, automate research workflows, and accelerate scientific discovery.”
While AI can help with tasks that are part of the scientific process, it is still far from automating science—and may never be able to. As a philosopher who studies both the history and conceptual foundations of science, I see several problems with the idea that AI systems can “do science” without or even better than humans.
AI models can only learn from human scientists
AI models don’t learn directly from the real world: They have to he “said” what the world was like by their human constructors. Without human scientists to oversee the construction of the digital “world” in which the model operates—that is, the datasets used to train and test its algorithms—the breakthroughs that AI facilitates would not be possible.
Consider the AI AlphaFold model. Its developers were awarded 2024 Nobel Prize in Chemistry for the model’s ability to infer the structure of proteins in human cells. Because so many biological functions depend on proteins, the ability to rapidly generate protein structures to test using simulations has the potential to speed drug design, track how diseases develop, and advance other areas of biomedical research.
As practical as it may be, an AI system like AlphaFold does not by itself provide new insights into proteins, diseases, or more effective drugs. It simply makes it possible to analyze existing information more effectively.
As philosopher Emily Sullivan has said, AI models must succeed as scientific tools maintain a strong empirical link to already established knowledge. This means that the model’s predictions must be based on what scientists already know about the natural world. The strength of this connection depends on how much knowledge is already available about the subject and how well the model’s programmers translate highly technical scientific concepts and logical principles into code.
AlphaFold wouldn’t be successful if it wasn’t the existing body of human knowledge about protein structures which the developers used to train the model. And without human scientists providing the foundation of theoretical and methodological knowledge, nothing AlphaFold creates would constitute scientific progress.
Science is a uniquely human enterprise
But the role of human scientists in the process of scientific discovery and experimentation goes beyond ensuring that AI models are properly designed and anchored to existing scientific knowledge. In a sense, science as a creative success derives its legitimacy from human abilityvalues and way of life. These, in turn, are based on the unique ways in which people think, feel and act.
Scientific discoveries are more than theories supported by evidence: they are the product of generations of scientists with diverse interests and perspectives, working together through a shared commitment to their craft and intellectual honesty. Scientific discoveries are never the product of a single visionary genius.
For example, when researchers first proposed double helix structure of DNAthere were no empirical tests to verify this hypothesis — it was based on the reasoning abilities of highly trained experts. It took nearly a century of technological progress and several generations of scientists to turn what seemed like pure speculation at the end of the 19th century into a discovery awarded the 1953 Nobel Prize.
In other words, science is a clearly a social enterprisein which ideas are discussed, interpretations are offered, and disagreements are not always overcome. As other philosophers of science have noted, scientists are more like a tribe than “passive recipients” of scientific information. Researchers do not gather scientific knowledge by recording ‘facts’ – they create scientific knowledge through skilled practice, debate and agreed standards based on social and political values.
AI is not a “scientist”
I believe that the computing power of artificial intelligence systems can be used to accelerate scientific progress, but only if it is done carefully.
With the active participation of the scientific community, ambitious projects like the Genesis mission could prove beneficial to scientists. Well-designed and rigorously trained AI tools would make the mechanical parts of scientific inquiry smoother and perhaps faster. These tools would gather information about what has been done in the past to more easily inform how to design future experiments, collect measurements, and formulate theories.
But if the main vision for deploying artificial intelligence models in science is to replace human scientists or fully automate the scientific process, I believe the project would only turn science into a caricature of itself. The very existence of science as a source of authoritative knowledge about the natural world depends fundamentally on human life: shared goals, experiences, and aspirations.
This edited article is republished from Conversation under a Creative Commons license. Read on original article.
Leave a Reply