How is AI used in life science research? ChatGPT, Claude and AlphaFold in preprints

6.5% of the preprints read so far say they used an AI assistant or language model, as a tool or as the thing the study tests, ChatGPT or another OpenAI GPT model most of all; 14.7% used an AI model built for science such as AlphaFold. Updated daily (as of Nov 16)

Counts preprints first posted Apr 2025–Nov 2026; Nov 2026 is still being read.

Since Apr 2025, ChatGPT is the AI assistant or language model used in the most preprints.

Claude comes next. In Oct 2026, the latest full month, Claude was used in more: 463 preprints to ChatGPT's 400. The counts are preprints that say they used each so far: 7,261 used an AI assistant or language model and 16,496 a science AI model.

AI assistants and language modelsPreprints

Use of AI assistants and language models has grown 4.4x since Apr 2025; science AI models are growing.

The share of each month's preprints saying they used an AI assistant or language model, or a science AI model, Apr 2025–Oct 2026, by the month each was first posted. AI assistants and language models: 3.0% of the preprints first posted in Apr 2025, 13.3% in Oct 2026; science AI models: 13.2% to 16.9%.

AI assistants and language modelsScience AI models

What do researchers use AI assistants and language models for?

From the sentence each preprint uses to say it used one (we never show the sentence). One preprint can state several uses.

UsePreprintsUsed most for itpreprints
Writing and editing2,979ChatGPT (1,766), Claude (820), Gemini (302)
Coding1,792Claude (791), ChatGPT (683), Claude Code (251)
Analyzing data1,406ChatGPT (570), Claude (345), Gemini (170)
Studying AI itself865ChatGPT (482), Claude (257), Gemini (226)
Figures and images309ChatGPT (127), Claude (58), Gemini (49)
Literature search251ChatGPT (98), Claude (51), Gemini (25)

Every number is a count of preprints. A preprint with several uses counts once for each, so the shares are of all uses stated; the other 16% are other uses, or the preprint doesn't say what for.

Which fields use AI assistants and language models most?

The share of each field's preprints that say they used one, as a tool or as the thing the study tests (fields with 500+ preprints read).

#FieldUsed or tested onePreprints read
1Health Informatics38.8% 6451,661
2Health Systems and Quality Improvement16.3% 94578
3Radiology and Imaging13.9% 102733
4Bioinformatics12.9% 1,0458,125
5Psychiatry and Clinical Psychology9.9% 1681,694
6Oncology9.8% 1041,062
7Systems Biology8.7% 1301,500
8Animal Behavior and Cognition7.3% 991,353
9Neurology7.1% 1702,378
10Epidemiology7.0% 1792,543
11Public and Global Health6.5% 1502,311
12Genetic and Genomic Medicine6.4% 1131,756

Least often: Immunology (2.7%), Biochemistry (3.0%), Developmental Biology (3.2%), Molecular Biology (3.3%), Microbiology (3.4%).

Rival tools, head to head

ChatGPT 3,963
AI assistants and language models
Claude 2,470

How does Tnuki know so much?

  1. Preprints analyzed dailyNew bioRxiv and medRxiv preprints are analyzed daily.
  2. Whole paper readWe assess the full text, tables, and supplementary files.
  3. Only real uses countedWe read to make sure the lab actually used the product, not just cited.
How we count

An AI model reads each bioRxiv and medRxiv preprint and lists what its authors used — instruments, kits, services, software, reagents — and how: used in this work, someone else's data, or only mentioned. A preprint counts here once, when the model says the lab used it, if it was first posted in Apr 2025 or later. Names are merged by a second model (versions and spellings into one tool, brands under their owner). We show counts and shares, and the titles of the latest preprints, each linking to the preprint on bioRxiv or medRxiv; we show no other text from a preprint and no author names. Updated daily, most recently on Nov 16, 2026.

Want to reach out to the labs using AI assistants?

Get the PIs behind the preprints: institutions, labs, and how to reach them.