In our posts dated February 18 and March 12, 2025, we explore how undergraduate STEM students integrate generative AI (GenAI) tools like ChatGPT, Gemini, and Perplexity into their academic studies routines. One major outcome our analysis revealed is that GenAI tool usage in science and engineering higher education evolves significantly as students progress through their academic journey. Specifically, as students advance, they become more comfortable integrating GenAI tools into their learning process, and their usage increases along several dimensions: frequency, number of usages, meta-cognition, technical capabilities, and academic skills.
In this post, we further analyze students’ usages of GenAI, exploring similarities and differences between two groups of undergraduate STEM students: those who are studying advanced computational (AC) fields such as computer science, electrical engineering, and data science, and all other undergraduate STEM students.
The Anthropic Education Report: How University Students Use Claude, published on April 8, 2025, presents findings based on the analysis of one million anonymized student conversations on Claude.ai. The report provides a broad overview of STEM students as early adopters of AI tools like Claude, with computer science students particularly over-represented, in contrast to business, healthcare, and humanities students, who exhibit lower adoption rates relative to their enrollment numbers. Our data analysis, presented in this post, deepens and complements the Anthropic Education Report observations in two ways: First, our analysis focuses only on STEM subjects and compares AC undergraduate students with all other STEM subjects undergraduate students; second, we track adoption rates among these two groups across four years of studies (freshmen, sophomores, juniors, and seniors), focusing on the evolution of GenAI usage over time.
Analyzing answers to our questionnaire (presented in the Appendix), we show that both the adoption rate and the use intensity of GenAI tools by AC students are higher than those of other STEM students.
Diffusion of GenAI Technologies: Undergraduate AC Students vs. Other Undergraduate STEM Students
Students in AC fields increasingly rely on GenAI as they progress in their academic education, as do other STEM students. Yet, their adoption rate is accelerated. While only 16% of first-year students in AC fields report using GenAI constantly, this figure increases to 43% by the fourth year. This is in contrast to students from other STEM disciplines, whose GenAI usage increases more modestly: from a comparable 15% constant use rate in the first year to 21% in the fourth year (see Figure 1). Concurrently, the percentage of AC students who never use GenAI drops from 14% in the first year to just 1% in the fourth, while the percentage of students from other disciplines who never use GenAI is 11% in both first and fourth year, with a little increase in the second and third year.
This data implies that AC students exhibit an accelerated diffusion of innovation rate (Rogers, 2003) compared with other STEM students; combined with the Anthropic Education Report’s observation, which characterizes all STEM students as early adopters of AI tools, our data implies that AC students can be characterized as accelerated early adopters.
To explore the nature of this accelerated diffusion of innovation, we present the distribution of GenAI tools usage among different categories (see Table 1). In this table, each academic year is presented separately and divided into AC and other STEM students.
| Freshmen | Sophomores | Juniors | Seniors | |||||
| AC fields | Other STEM subjects | AC fields | Other STEM subjects | AC fields | Other STEM subjects | AC fields | `Other STEM subjects | |
| N | 116 | 157 | 90 | 133 | 62 | 86 | 80 | 102 |
| Understanding course material | 66% | 49% | 52% | 48% | 37% | 43% | 48% | 33% |
| Help with homework | 33% | 54% | 40% | 42% | 28% | 29% | 31% | 22% |
| Programming | 9% | 2% | 22% | 8% | 37% | 21% | 44% | 24% |
| Study materials | 13% | 11% | 5% | 11% | 9% | 21% | 14% | 18% |
| Writing | 2% | 1% | 0% | 7% | 5% | 14% | 20% | 22% |
| Academic papers | 0% | 0% | 0% | 0% | 2% | 10% | 5% | 10% |
| Deepening knowledge | 2% | 0% | 7% | 6% | 5% | 3% | 5% | 6% |
| Exam preparation | 2% | 0% | 5% | 3% | 0% | 2% | 2% | 3% |
| Advanced academic activities | 2% | 1% | 2% | 2% | 2% | 8% | 5% | 3% |
According to the distribution presented in Table 1, the nature of GenAI usage among AC students differs from that of other STEM students in two ways:
a) Supportive role of GenAI tools. The percentage of AC students that use GenAI to understand course material is 66% for freshmen and 48% for seniors, compared with 49% and 33% for freshmen and seniors of other disciplines, respectively. This can be explained by the more demanding learning material in AC fields.
b) GenAI tools as practical resources. By the third and fourth years, programming becomes a key use case, with 37% of AC juniors and 44% of AC seniors using GenAI for this purpose. Other STEM students exhibit different patterns. Freshmen use GenAI significantly less for programming (2% vs. 9%), and by the fourth year, the gap in programming-related use widens even further (24% vs. 44%). This evolution reflects the code-oriented nature of AC studies, an area in which GenAI tools excel, and suggests that for AC students, GenAI tools are not simply scaffolds for foundational learning, but rather resources that AC juniors and seniors learn to use over time.
These contrasting patterns reinforce the claim that AC students may be acting as accelerated early adopters not only in terms of frequency of use, but also in terms of the depth and complexity of their use of GenAI applications. This assertion stems from a combination of more demanding learning material and the code-oriented nature of the disciplines. The students’ increasing reliance on GenAI over time demonstrates how innovations diffuse more rapidly among populations motivated to explore them for practical purposes.
Implications for Teaching and Curriculum Design
We conclude with several pedagogical insights for teaching in AC fields, as well as in other STEM disciplines:
Since AC students already present an accelerated rate of adopting GenAI tools for use in their learning process, instructors can harness this usage to enhance learning. For example, they can reframe learning tasks to encourage students to use GenAI in ways that emphasize debugging, code comprehension, or computational thinking.
The patterns observed in accelerated-adoption groups, from being passive consumers of answers to being active, can serve as a roadmap for creating scaffolded experiences that gradually build confidence and competence with GenAI tools in other STEM fields, as well as in all other disciplines. By understanding how these students adopt GenAI, we can design educational experiences that prepare all learners to overcome adoption barriers on their track to adopt intelligent technologies.
Rather than restricting GenAI, we suggest that teachers foster digital literacy and critical engagement with it. Educators can build on their students’ growing familiarity with GenAI by designing tasks that encourage collaborative learning with AI partners while reflecting on their work with these tools.
References
Rogers, E. (2003). Diffusion of Innovations, 5th Edition. Simon and Schuster. ISBN 978-0-7432-5823-4.
Appendix
Each year, during the fifth week of the Technion’s winter semester, a survey is distributed to all undergraduates (about 10,000 students in total). The purpose of the survey is to let the students’ voices and perspectives regarding their learning experience at the Technion be heard, and to take relevant action according to the survey findings. This academic year (2024-2025), two new questions were added to the survey that address the use of GenAI tools by Technion students. The purpose of these additional questions was to discover the extent and ways in which students use GenAI tools spontaneously, and to derive insights about science and engineering learning and teaching using GenAI.
Yael Erez is a lecturer at the Technion’s Faculty of Computer Science and a faculty member at the Department of Electrical Engineering at the Braude College of Engineering in Karmiel, Israel. She is currently a doctoral student at the Technion’s Department of Education in Science and Technology, under the supervision of Orit Hazzan.
Orit Hazzan is a professor at the Technion’s Department of Education in Science and Technology. Her research focuses on computer science, software engineering, and data science education. See https://orithazzan.net.technion.ac.il/.
