# AI usage guide for labs: 4 distinct approaches Discover how to implement AI based on your research values - **Source:** [Nature](https://brifing.si/source/4d850c3b-dd79-4024-bc91-db2571889199) - **Published:** 2026-08-31 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/b1b16ef6-ab26-48f5-8c5b-49d8be863f0c ## Unique approach to AI in science A new paper published on arXiv reveals there's no one-size-fits-all solution for lab AI adoption. Astronomer Sarah Burke-Spolaor from West Virginia University and her team demonstrate how research groups with different values might utilize AI differently. Although not yet peer-reviewed as of July 2026, the study has generated significant interest in the scientific community. ## Four laboratory archetypes Researchers categorize labs into four fundamental types: 'High Leverage', 'Craftsmanship', 'Trustworthiness', and 'Data Stewardship'. Each type has distinct AI usage priorities. For instance, resource-conscious high leverage labs employ AI for code generation and data analysis, while transparency-focused trustworthiness teams prefer using it primarily for debugging purposes. ## High leverage laboratories These teams aim to maximize scientific impact with limited resources. They actively use large language models in code generation, data analysis, and brainstorming processes. Michelle Ntampaka from Maryland's Space Telescope Science Institute notes this approach proves particularly advantageous for fast-moving projects requiring rapid progress. ## Craftsmanship labs Groups prioritizing expertise development and methodical work primarily utilize AI for administrative tasks like writing assistance and meeting summaries. Harvard Medical School's Tari Tan emphasizes these teams avoid delegating 'hard thinking' processes to AI, valuing the preservation of fundamental research skills. ## Trustworthiness-focused teams Labs where reproducibility and transparency are critical mainly employ AI for code debugging and documentation processes. Burke-Spolaor highlights this approach effectively maintains data integrity in fields like astronomy that handle massive datasets. ## Data stewardship groups Teams working with sensitive data typically prefer self-hosted AI models, avoiding cloud-based solutions due to privacy concerns. Ntampaka notes these groups prioritize data responsibility and approach AI integration with greater caution compared to other archetypes. ## Hybrid approaches and flexibility Researchers stress most labs may adopt combinations of these categories. Priorities can shift based on project type and phase - for example, high leverage approaches during rapid development stages versus craftsmanship during phases requiring methodological rigor. This fluidity allows labs to adapt their AI usage as research needs evolve. ## Roadmap for AI policies The paper provides practical worksheets to help teams develop customized AI strategies. Burke-Spolaor suggests research leaders begin by asking 'what kind of team do we want to sculpt?' This approach enables each lab to create solutions tailored to their specific needs, independent of institutional policies. ## Sources covering this story - https://www.nature.com/articles/d41586-026-02543-z — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._