SARSEF’s high school research and mentorship program, PRISM, recently kicked off a second year in August 2026. At the orientation event, students connected with their cohort of peers and “matched” with a mentor for the semester, with whom they will be working on an original research project. Students and mentors also worked together to create a shared understanding and set of rules for the use of artificial intelligence (AI) in a research framework.
As different forms of AI (notably chatbots such as ChatGPT and Claude) have become increasingly prevalent in how people interact with technology in their daily lives, a major question remains about where AI belongs in education for youth. Through the PRISM program, Brianne Loya, High School Education Director at SARSEF, asked the question of “what is the place of AI in research” to the 2026 PRISM cohort.
At this point in AI’s evolution, it seems that every school, if not every classroom, is in charge of creating guidelines for their students on the appropriate use of AI in school. There are no nationally recognized frameworks for how AI should be treated in education. Similarly, few frameworks exist that students, mentors, and other professionals can look to in order to make sense of the potential role of AI in research. The SARSEF Fair has implemented a guideline for the use of AI in science fair projects, which are based on national models from the Society for Science and other institutions. These guidelines are specific for science fair projects and appeal to a range of grade levels and experiences. The guidelines encourage students to think critically about the use of AI in science fair project development, asking students to evaluate the strengths and weaknesses that current AI technologies have. However, the guidelines are not binding agreements and allow for flexibility of interpretation. The problem remained: when students are researching professionally, such as through the PRISM program, how should AI be treated?
The process of creating the shared understanding and rules for the PRISM program began with a set of questions: (1) What do we believe learning is for ourselves as students? (2) What do we broadly believe about technology, including AI? These questions helped develop the following discussion and activity about how exactly AI can be used in student research. Throughout the discussion, ideas started to pop up about technology, such as that technology is not a neutral tool and that it influences how students think, communicate, and form ideas today. A popular example that emerged is how students can use AI for writing, and whether a student is actually learning any content or skills when using AI in specific ways for writing. Students, with the help of their mentors, began to think through how AI influences learning—not through whether AI can successfully complete a research task sufficiently—and whether AI can actually help people learn or hinder critical thinking.
Students and mentors were then grouped and tasked with sorting a set of scenarios about the use of AI in the PRISM program on cards into three categories: acceptable in PRISM, not acceptable as described but may be if revised, or not acceptable in PRISM. Scenarios included asking students whether it is acceptable to have AI summarize research articles without engaging with the original content, generate research questions for the student, use AI to modify procedures, or use AI to edit writing.
Every group came up with a slightly different arrangement of the scenarios between the three categories. Before a full team discussion, groups were asked to analyze the scenarios they put in the middle category of “not acceptable as described” and then either revise them to make them acceptable or ultimately move them to the not acceptable pile. Teams then discussed with the whole group how they made decisions about using AI in research, what card placements they disagreed with each other about, and whether there were principles about AI use that could be applied to other scenarios not specifically covered in the activity.
Ultimately, the activity yielded a functional AI Use Policy for the PRISM program, where acceptable and unacceptable usage of AI in research is clearly outlined. The whole group also determined that any use of AI, even if it is for grammatical editing, needs to be documented in laboratory notebooks and have a statement about how AI was used in any relevant documents, charts, or figures, if applicable. Through the activity, students and mentors together critically analyzed the use of AI and asked themselves, “Is this how I want to use AI?” considering the shared understanding of what learning means.
“We want our students to be able to make informed decisions about new technology,” says Brianne Loya. She came up with the scenarios used in the activity based on experiences and uses she encountered during her time as a science teacher in the Phoenix area.
The AI Use Policy is a resource, and each student and mentor has agency over whether or not they use AI in the first place. The policy is also a living document, where year-over-year, each cohort of students and mentors will be asked to form their own policy that reflects the nature of AI in their situation.
Written by Claire Gibson, Marketing and Communications Manager, SARSEF