A recent study by the University of Chicago and Argonne National Laboratory examines generative AI\u00a0tools \u2014 specifically\u00a0large language models (LLMs) \u2014 within a national lab setting. The study not only highlights AI\u2019s potential to enhance productivity, but also emphasizes the need to address such areas such as privacy, security and transparency.<\/p>\n
Through surveys and interviews, the researchers studied how Argonne uses LLMs \u2014 and their future use \u2014 to generate content and automate workflows. The study also tracked the early adoption of Argo, the lab\u2019s internal LLM interface released in 2024.<\/p>\n
Argonne and Argo<\/strong><\/p>\n Argonne said its organizational structure paired with Argo made the lab an ideal environment for the study. Its workforce includes science and engineering workers as well as employees in such areas as human resources, facilities and finance.<\/p>\n Kelly Wagman<\/p>\n The researchers found that employees primarily used genAI as a copilot and as a workflow agent. As a copilot, the\u00a0AI\u00a0works alongside the user, helping with tasks like writing code, structuring text or tweaking the tone of an email. For the most part, employees are currently sticking to tasks where they can easily check the\u00a0AI\u2019s work. In the future, employees reported envisioning using copilots to extract insights from large amounts of text, such as scientific literature or survey data.<\/p>\n As a workflow agent,\u00a0AI\u00a0is used to automate complex tasks, which it performs mostly on its own. Around a quarter of the survey\u2019s open-ended responses \u2014 split evenly between operations and science workers \u2014 mentioned workflow automation, but the types of workflows differed between the two groups. For example, operations workers used\u00a0AI\u00a0to automate processes like searching databases or tracking projects. Scientists reported automating workflows for processing, analyzing and visualizing data.<\/p>\n \u201cScience is an area where human-machine collaboration can lead to significant breakthroughs for society,\u201d said Kelly Wagman<\/a>, a Ph.D. student in computer science at the UChicago and lead author on the study.\u00a0\u200b\u201cBoth science and operations workers are crucial to the success of a laboratory, so we wanted to explore how each group engages with\u00a0AI\u00a0and where their needs align and diverge.\u201d<\/p>\n While the study focused on a national lab, some of the findings can extend to other organizations, including universities, law firms and banks, which have varied user needs and similar cybersecurity challenges.<\/p>\n Argonne employees regularly work with sensitive data, including unpublished scientific results, controlled unclassified documents and proprietary information. In 2024, the lab launched Argo, which gives employees secure access to LLMs from OpenAI through an internal interface. Argo doesn\u2019t store or share user data, which Argonne said makes it a more secure alternative to ChatGPT and other commercial tools.<\/p>\n Argo was the first internal genAI interface to be deployed at a national laboratory. For several months after Argo\u2019s launch, the researchers tracked how it was used across the lab. Analysis revealed a small but growing user base of both science and operations workers.<\/p>\n Possibilities and Risks<\/strong><\/p>\n While generative\u00a0AI\u00a0presents exciting opportunities, the researchers also emphasize the importance of thoughtful integration of these tools to manage organizational risks and address employee concerns.<\/p>\n The study found that employees were significantly concerned about generative\u00a0AI\u2019s reliability and its tendency to hallucinate. Other concerns included data privacy and security, overreliance on\u00a0AI, potential impacts on hiring and implications for scientific publishing and citation.<\/p>\n To promote the appropriate use of generative\u00a0AI, the researchers recommend that organizations proactively manage security risks, set clear policies and offer employee training.<\/p>\n![\"\"](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2025\/04\/kelly-wagman-university-chicago-0425.jpg\")
<\/p>\n
![\"\"](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2025\/04\/argonne-logo-2-1-04-14-25.png\"){kind=logo}
\u201cWithout clear guidelines, there will be a lot of variability in what people think is acceptable,\u201d said Marshini Chetty, a professor of computer science and leader of the\u00a0