AI tools now summarize literature, write analysis code, suggest hypotheses and draft papers. That leads people to ask whether scientists themselves will be needed. The data points the other way: science jobs score low to moderate for generative AI exposure, and US projections expect most science fields to grow.
The short answer: current evidence does not point to AI replacing scientists. The ILO's 2025 index places most science occupations in its minimal exposure band or low gradients. The US Bureau of Labor Statistics projects growth from 2025 to 2035 for medical scientists, chemists and materials scientists, environmental scientists and physicists and astronomers. AI is speeding up reading, coding and writing; designing experiments, judging evidence and lab and field work stay human. Projections can change.
How exposed science jobs are
From the ILO's 2025 index, which scores tasks 0 to 1 for generative AI exposure:
| Occupation (ISCO-08) | ILO band | Mean task score |
|---|---|---|
| Biologists, Botanists, Zoologists and Related Professionals | Gradient 2 | 0.40 |
| Chemists | Minimal exposure | 0.39 |
| Physicists and Astronomers | Gradient 1 | 0.38 |
| Life Science Technicians (excluding Medical) | Gradient 1 | 0.38 |
| Environmental Protection Professionals | Minimal exposure | 0.38 |
| Geologists and Geophysicists | Minimal exposure | 0.36 |
Gradient 1 means low overall exposure with some highly exposed tasks, such as writing up results. Lab and field work keep overall scores down.
What the BLS projects
| Occupation | Projected change 2025 to 2035 | Openings a year |
|---|---|---|
| Medical scientists | Grow 13% | About 10,500 |
| Chemists and materials scientists | Grow 7% | About 6,500 |
| Physicists and astronomers | Grow 7% | About 1,500 |
| Environmental scientists and specialists | Grow 6% | About 7,300 |
The BLS links medical scientist demand to an aging population and rising chronic disease, and research into diseases such as Alzheimer's and cancer and into antibiotic resistance. For chemists and materials scientists it points to innovation in areas such as nanotechnology and biomaterials; for environmental scientists, public interest in environmental hazards and business demand for advice; for physicists and astronomers, deep-space missions, while noting that federal spending is a primary source of research funds and the federal budget may directly affect employment.
Research tasks AI is taking on
- Searching and summarizing the literature.
- Writing and debugging analysis code.
- Drafting methods sections, grant text and reports.
- Cleaning, sorting and visualizing data.
- Suggesting candidate hypotheses or designs to test.
What stays with the scientist
- Asking the right question and designing a study that can answer it.
- Lab and field work: running experiments, handling samples, fieldwork.
- Judging evidence: whether a result is real, reproducible and meaningful.
- Checking AI output: AI can produce confident but wrong statements and citations to papers that do not exist, which NIST calls confabulation (NIST AI 600-1). Every citation and number needs verifying. See checking AI output.
- Validity of models: the NIST AI RMF lists valid and reliable among the characteristics of trustworthy AI (NIST AI RMF 1.0), and scientists using AI models must test that for their own data.
- Ethics, peer review and responsibility for published work.
How the job is changing
For scientists, AI is becoming a research assistant for reading, coding and drafting, and a tool for analyzing large datasets. That can mean more output per researcher and new kinds of questions. Funding remains one of the main factors the BLS names for jobs in several fields. Scientists who combine domain depth with careful use and testing of AI tools have the advantage.
See also will AI replace data analysts?, will AI replace engineers? and will AI take my job?
Pay and how to get in
The US Bureau of Labor Statistics puts the median annual pay for medical scientists at $103,410 in May 2025, against $50,980 for all workers. Across the 4 BLS occupations this page covers, median pay ranges from $82,220 to $166,880. The typical education needed to enter differs between the 4 BLS occupations this page covers. See how this job compares in AI exposure by pay and AI exposure by education level.
Questions and answers
Will AI replace scientists?
Not on current evidence. Most science occupations are in the ILO minimal exposure band or low gradients, and the BLS projects 6 to 13 percent US growth from 2025 to 2035 for medical, chemical, environmental and physical scientists.
How exposed are scientists to AI?
In the ILO 2025 index, science occupations score between about 0.36 and 0.40, mostly in the minimal exposure band or gradients 1 and 2.
How do scientists use AI?
For literature searches and summaries, analysis code, drafting methods and reports, data cleaning and visualization, and suggesting hypotheses to test.
What research work stays human?
Choosing questions, designing studies, lab and field work, judging evidence, checking AI output and citations, testing model validity and taking responsibility for published work.
Is science still a good career?
US projections expect growth in most science fields, with federal research budgets an important factor in some, such as physics and astronomy.