Generative AI Usage Patterns and User Profiles Among Researchers: Evidence from a Large-Scale Survey in China
This study utilizes large-scale survey data from Chinese researchers to extend the Unified Theory of Acceptance and Use of Technology with machine learning, identifying three distinct GenAI user profiles—"Pioneering Explorers," "Pragmatic Coping Users," and "Alienated Users"—that reveal how academic standing, institutional pressures, and ethical concerns shape divergent adoption patterns in the context of AI for Science.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the vast landscape of modern science, a new tool has emerged that promises to reshape how discoveries are made. This tool is generative artificial intelligence, a type of computer program capable of creating text, images, and code that mimics human creativity. Unlike earlier computers that simply followed strict instructions, these systems can hold conversations, summarize complex ideas, and even suggest new research directions. For scientists, this technology offers the tantalizing possibility of accelerating the slow, often tedious work of reading papers, writing reports, and analyzing data. However, just as with any powerful new invention, the way people interact with it is not uniform. Some embrace it fully, while others remain cautious or skeptical. Understanding why these differences exist is crucial, because the future of scientific progress may depend on how well researchers can integrate these tools into their daily lives without losing sight of accuracy, ethics, or the human element of discovery.
A large-scale study conducted in China has taken a deep dive into this very question, mapping out exactly how researchers are using generative artificial intelligence and what drives their choices. The researchers surveyed nearly 18,500 academic scientists across the country, eventually focusing their detailed analysis on a group of 2,507 who had already used the technology. Instead of simply asking whether people liked the tool or how often they used it, the team looked at the specific tasks these scientists performed, their personal backgrounds, their fears, and their daily work habits. By analyzing this massive amount of information, they discovered that researchers do not fall into a single category of "user" or "non-user." Instead, they naturally cluster into three distinct groups, each with its own unique profile, motivations, and relationship with the technology.
The first group, which the researchers call "Pioneering Explorers," represents about 19 percent of the users. These are scientists who use the technology frequently and for a wide variety of tasks, ranging from writing papers to designing experiments and training complex computer models. These individuals tend to hold senior positions within the scientific community, possessing high academic standing and exceptional work engagement. They often come from families with high levels of education and earn higher incomes. What sets them apart is their mindset: they are deeply aware of the technology's flaws, such as its tendency to make up facts or lack logical depth, yet they are driven by the sheer potential of the tool to automate difficult parts of their work. They see the technology not just as a writing assistant, but as a partner in the entire scientific process, from gathering data to testing new hypotheses. Their high level of engagement with their work and their diverse professional networks seem to fuel this broad, confident exploration.
In contrast, the second group, dubbed "Pragmatic Coping Users," makes up the largest slice of the population at nearly 47 percent. These are typically younger researchers working in wealthy regions or at elite universities. They face intense pressure to publish papers quickly and often struggle with fragmented, busy schedules. For them, the technology is a lifeline to manage this workload. They use it frequently, but almost exclusively for reading literature and writing papers. They are less concerned with the technology's ability to help with deep scientific discovery and more focused on its ability to speed up the publication process. Interestingly, this group tends to underestimate the risks of the technology, particularly regarding intellectual property and logical errors. They view the tool as a practical solution to immediate academic stress, using it to clear the path for their next paper rather than to fundamentally change how they do science.
The third group, known as "Alienated Users," accounts for about 35 percent of the sample. These researchers use the technology very rarely, if at all. They are generally older, often over the age of 45, and work in basic sciences like mathematics or physics, or in roles that involve significant teaching responsibilities. They tend to have lower publication outputs and lower self-rated professional standing. They hold deep reservations about the technology, fearing its ethical implications and doubting its ability to produce accurate or meaningful results. They do not see the technology as a threat to their jobs, nor do they believe it can automate complex scientific thinking. Because they have not engaged deeply with the tools, they remain skeptical of their capabilities, creating a cycle where their lack of use prevents them from seeing the potential benefits that might change their minds.
The study reveals that the decision to use generative artificial intelligence is not merely a technical choice but a reflection of a researcher's entire professional life. For the "Pioneering Explorers," the drive to innovate and the resources to manage risk allow them to embrace the technology fully. For the "Pragmatic Coping Users," the pressure of the academic system pushes them to use the tool as a survival mechanism, focusing narrowly on efficiency. For the "Alienated Users," a combination of age, field of study, and deep-seated concerns keeps them on the sidelines. The researchers found that perceived risks do not automatically stop people from using the technology; rather, it is a balance between those risks and the perceived benefits. Those who see the greatest potential for automation and efficiency are willing to overlook the flaws, while those who see only the dangers or no clear benefit stay away.
Ultimately, this research suggests that the future of science with artificial intelligence will not be a uniform wave of adoption. Instead, it will be shaped by the diverse realities of the scientists themselves. The study highlights that for the technology to truly transform science, the academic environment itself may need to change. If researchers are constantly rushed and fragmented by excessive publication demands, they may only use these powerful tools for quick fixes, missing out on the deeper, more revolutionary possibilities. Conversely, those with the freedom and resources to explore deeply are already pushing the boundaries. The findings serve as a snapshot of a critical moment in time, showing that the path forward for artificial intelligence in science depends as much on the people using it as on the technology itself.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.