Guidance for the Optional Use of Artificial Intelligence (AI) in Summarizing Student Feedback for Faculty Assessment

This guidance establishes best practices for optionally using AI to summarize qualitative student feedback as one component of the assessment of teaching for faculty members (tenure-line and non-tenure-line) undergoing formal review. AI must not be used to generate evaluative statements about the candidate’s teaching effectiveness. These judgments, opinions, and perceptions must rely solely on the human insight and discernment of reviewers (e.g., the two individuals serving as student feedback reviewers selected per the Changes to Assessment of Teaching Effectiveness, hereafter referred to as “reviewers”). AI should only be used to summarize large volumes of qualitative data (e.g., student comments) to inform the overall assessment of a candidate’s teaching effectiveness. Adhering to these best practices helps ensure consistency and transparency, thereby mitigating bias within and across units. 

Beginning July 1, 2025, all colleges are expected to integrate overarching principles for the incorporation of student feedback into promotion and/or tenure guidelines for tenure-line and non-tenure-line faculty members undergoing formal review (see Changes to Assessment of Teaching Effectiveness). Reviewers in this process are charged with: (1) examining student feedback from available courses covering the period since a candidate’s last formal review or the review period (whichever is shorter), and (2) writing a report of no more than 750 words (approximately one single-spaced page) describing their insights into the candidate’s teaching effectiveness based on quantitative and qualitative student feedback from SEEQ/SRTE responses across the review period. 

Reviewers who encounter a large number of qualitative student comments may optionally choose to use AI for an initial summary. Reviewers can independently determine if the volume (e.g., number of student comments) and/or quality (e.g., detailed nature of the student comments) justifies the use of AI to manage the review workload. However, AI is only permitted to summarize comments and cannot be used to evaluate student feedback or provide insight into the candidate’s teaching effectiveness. These evaluations must be made through the careful judgment of the reviewers. 

It is important to note that the required 750-word report in the assessment of teaching process is an evaluation of the candidate’s teaching effectiveness, based on an analysis of students’ assessments regarding the candidate’s performance. It is a synthesis and interpretation of both quantitative and qualitative feedback from SEEQ/SRTE responses across all courses taught during the review period (see Changes to Assessment of Teaching Effectiveness). It is not a summary. The use of AI to summarize qualitative student feedback is but one part of the data that informs this review. Again, under no circumstances should AI be used to provide an evaluation of the student comments, the candidate’s teaching effectiveness, or write the final report. 

Reviewers who choose to use AI for the purposes of summarizing qualitative student feedback must: 

• disclose the use of AI in summarizing qualitative SEEQ/SRTE student feedback via a footnote in the final report. The disclosure must include the AI platform used, the exact prompt, and the number of qualitative comments analyzed (see additional guidance below). The word count for this disclosure will not be included as part of the 750-word limit for the final report. 

• use Microsoft Copilot as the platform to run the summary analysis. Penn State has contracted with Microsoft to ensure Copilot maintains security and confidentiality and does not use entered data for training. Access is restricted to Penn State-authenticated users. 

• only include open-ended SEEQ responses A1 and A4 in the AI analysis. Quantitative data A2 and A3 should not be included in the AI analysis. Reviewers may not request candidates to share student feedback intended for only the instructor, which includes results from the MSEEQ and items A5, A6, and A7 of the SEEQ (see SEEQ Items). All SRTE student comments from the review period must be included. 

• remove any personally identifiable information (e.g., any text that identifies a specific individual) found in the comments prior to entering the student comments into the AI (note that Penn State’s contracted Microsoft Copilot covers level 1 and 2 information classifications). 

• use the following prompt to generate the summary: “Please take on the role of a university professor. Please provide an overall summary of the student evaluations that I will provide to you, including key themes, sentiment analysis, and any notable trends or patterns. Do you understand?” Reviewers should not perform any additional AI analysis of the student comments beyond this prompt or those suggested by AI as a result of this prompt. 

• review the AI summary of qualitative student comments, along with other sources of student feedback, and write the required 750-word report of the candidate’s teaching effectiveness. This is not a summary. The report must describe the candidate’s teaching effectiveness, must be written with the discernment and judgment of the reviewers, and must be based on the review of both the quantitative and qualitative SEEQ/SRTE student feedback. 

Please note that all aspects of this process are confidential. All reviewed materials and notes, including the AI output, should be destroyed immediately after the final report is submitted. 

Per the Assessment of Teaching guidelines, if a candidate believes the final report does not accurately reflect their teaching effectiveness, they may revise their narrative to address the discrepancy. 

Please see Frequently Asked Questions for Administrators about Student Feedback Use in the Assessment of Teaching Effectiveness for more information.