Artificial intelligence is rapidly reshaping the scientific workforce, with several roles already showing signs of strain.
In a wide-ranging report, Nature spoke with more than 40 researchers from academia and industry who actively use AI in their work. Many pointed to a noticeable decline in demand for scientists զբաղing in routine coding and basic data analysis.
These tasks have traditionally been carried out by PhD candidates, postdoctoral researchers, or technical staff without senior research titles. Increasingly, however, AI systems are performing them faster and at lower cost. As MIT mechanical engineer Xuanhe Zhao noted, the erosion of certain roles—particularly in computational modeling—is “not a future concern; it is happening now.”
Spillover effects beyond core research
The impact extends beyond frontline research positions. Adjacent professions, such as scientific translation, are also being affected. With rapid advances in AI-powered translation tools, parts of the field are shrinking quickly. The American Translators Association, for instance, reported a 26% drop in membership within its Science and Technology division in under two and a half years, prompting some professionals to change careers altogether.
Coding and data roles most exposed
According to many researchers, the most immediate disruption is concentrated in coding and data processing. Laboratories that once relied on dedicated programmers to build software tools are now reconsidering those hires. Computational biologist Hannah Wayment-Steele says that while such roles were once essential, AI can now handle even complex programming tasks, reducing the need for specialized staff.
Fewer entry points for young scientists
Hiring trends are also shifting. Many labs are becoming more cautious about recruiting graduate students and postdoctoral researchers, citing both funding uncertainty and the growing capabilities of AI.
Even without widespread layoffs, the slowdown in new positions is raising concerns about the long-term pipeline of scientific talent. Early-career roles, often critical stepping stones, may become harder to access.
The human factor remains critical
Despite its rapid progress, AI still has clear limitations. While it can summarize literature, refine manuscripts, generate code, and process data, it has yet to demonstrate the ability to consistently produce genuinely original ideas or groundbreaking research questions.
Physicist Jonathan Oppenheim notes that although he uses AI to generate draft peer-review reports, the systems fall short when it comes to true innovation. Similarly, computer scientist Karu Sankaralingam argues that the most productive path forward lies in combining human insight with machine capabilities, keeping a “human in the loop” to guide outputs and avoid errors.
Where jobs remain safer
Roles involving experimental and hands-on laboratory work appear more resilient for now. Despite advances in automation, many complex procedures still require human dexterity and judgment.
Structural biology offers a telling example: even though AI tools like AlphaFold2 have transformed protein structure prediction, labor-intensive experimental techniques remain indispensable, particularly in challenging cases.
Researchers suggest that adaptability will be crucial for the future of science. As mathematician Terence Tao puts it, those who successfully integrate AI into their workflows may not only endure—but potentially thrive.