Roundup on Research: Examining Feedback, GenAI, and Instructor Presence

Many instructors have long grappled with how to provide meaningful feedback at scale, particularly in online courses. Generative AI has intensified this challenge, not by creating a new problem, but by making long-standing tensions around feedback more visible. As feedback becomes easier to generate, the role of the instructor becomes more consequential, shifting from provider of comments to designer of feedback systems.

While feedback is often framed as information about students’ work, decades of research suggest it has never been purely informational (Evans, 2013). The amount of detail, tone, timing, modality, and perceived authority of feedback all shape how students interpret and act on instructor guidance. Recent research, including emerging work on generative AI, reinforces that feedback is a multidimensional instructional practice rather than a single act.

Providing meaningful, timely, and actionable feedback is both a cornerstone of education and also a challenge. Teaching has always required instructors to balance scale, presence, and personalization under tight deadlines. With the advent of more automated tools like GenAI, the question inevitably arises: is instructor feedback still useful? Recent research suggests that what GenAI changes is not the importance of feedback, but how visible the design decisions behind it have become and, in particular, decisions about where instructor judgment matters most.

Here we review recent articles on classroom feedback, including how GenAI can help, and where the instructor remains most important.

What We Know About Feedback

Across recent research, the strongest consensus is that there is no single “best” form of feedback. Common assumptions such as feedback must be written, feedback must come exclusively from the instructor, include extensive detail, or remain emotionally neutral are not supported by the evidence. Instead, feedback effectiveness depends on how multiple dimensions align with instructional goals.

Schluer and Brück-Hübner (2025) identify seven dimensions of pedagogical feedback, including source, purpose, timing, and modality. As none of those dimensions is more important than the others, the study suggests that effective feedback is a nuanced process that still requires sustained instructor engagement. 

Other types of feedback are still valuable though. Both Lopera-Oquendo and colleagues (2025) as well as Li & Collins (2026) found that feedback from instructors was still the most valued. Even though students report value in feedback from other sources like peers, and even GenAI, it is the instructor voice that holds the most weight. This places additional pressure on faculty as feedback is time-intensive, and the scale of online courses continues to grow.

What GenAI Changes, and What it Doesn’t

Recent research suggests that generative AI is particularly well suited to identifying where feedback may be needed in student work. In Kashiha’s (2025) study, both instructors and ChatGPT identified similar areas in student writing that warranted feedback. However, the feedback itself differed substantially. AI-generated feedback tended to be broader and more generalized, while instructor feedback was more targeted, directive, and strategically prioritized.

This distinction highlights an important design opportunity. GenAI can function as a highlighter, surfacing patterns or common issues across student work, while instructors retain responsibility for interpreting those issues, like deciding which matter most, how to frame next steps, and how to align feedback with course goals.

GenAI can also generate feedback at scale, which may be tempting in high-enrollment or online courses. However, more feedback is not necessarily better feedback. Adıgüzel and colleagues (2025) suggest that overly detailed or excessive feedback can increase cognitive load and anxiety, particularly in online environments. Targeted, actionable feedback delivered with clear instructional intent remains more effective than volume alone.

One final place where GenAI can provide meaningful feedback to students is as an intelligent tutor (Wang and Fan, 2025). In this role, GenAI functions less as an evaluator and more as a formative feedback mechanism, helping students recognize gaps in understanding before formal assessment. If students are struggling with complex topics, GenAI can help break down complex constructs in a way students can understand more easily, providing feedback on the content. Wang and Fan found the best results occurred when GenAI is integrated over a longer period of time, between four and eight weeks.

Instructors Matter More, Not Less

Across these recent studies, instructor feedback is consistently perceived by students as the most valuable. While students recognize the usefulness of peer and AI-generated feedback for specific purposes, instructor feedback carries unique authority. It signals prioritization, establishes expectations, and connects individual performance to broader course goals.

For example, in Sandidge & Schultz’s (2024) study examining the use of discussion boards in the online classroom, the researchers surveyed students about their views on the discussion board model (one post, two replies to other students). While students generally viewed the discussion board practice unfavorably, they found discussion board activities valuable when the instructor participated. Instructors guide, prioritize, and humanize material.

Human instructors also convey presence through feedback. First-person language, personalized references, and explicit guidance communicate that a knowledgeable instructor is engaged in the learning process. This combination of authority and presence helps students interpret feedback not merely as correction, but as coaching.

A clear distinction emerges across this newest literature between identifying where feedback is needed and interpreting what that feedback should mean for learning. Generative AI appears well-suited to the former, while instructors remain essential for the latter.

Practical Implications

The goal of using GenAI is not to remove the instructor from the process, but to shift your role from a “provider of comments” to a “designer of feedback.” Instead of spending time hunting for the same repetitive errors, you can use technology to help you focus on where your expertise matters most.

Use GenAI as a Highlighter, not a Grader

Research shows that GenAI and instructors often identify the same areas that need improvement. You can use this to your advantage to save time:

• Identify patterns: Use GenAI to find common writing errors or recurring themes across all student submissions.
• Focus on interpretation: Let the AI handle the “identification” of the problem, so you can provide the “interpretation” of explaining why it matters and how to fix it.

• Prioritize growth: Rather than providing a mountain of corrections that can cause student anxiety, use GenAI to surface issues and then use your judgment to prioritize the two or three most important changes for the student.

Lean Into Your Human Presence

Students value instructor feedback above all other sources because of your authority and coaching presence. To maintain this connection:

• Use first-person language: Insert your presence by using “I” statements (e.g., “I suggest” or “I noticed”) to signal that you are coaching them, not just correcting them.
• Be strategic, not universal: While AI provides “softer” or more universal feedback, your feedback should be direct and actionable.
• Experiment with video: If writing detailed comments feels overwhelming, try video feedback. It has shown a large impact on student learning and helps convey a supportive tone that text often lacks.

Build a Long-Term Collaboration

Simply using GenAI for a single assignment rarely provides significant learning benefits. For the best results:

• Let GenAI offer formative feedback to students on complex topics.
• Integrate over time: Think of GenAI as an “intelligent tutor” that students use consistently over four to eight weeks.
• Guide thinking: Use the tool to help students break down complex ideas, which helps them focus their own creativity and critical thinking.

References

Adıgüzel, O. C., Esen, E., & Karagöl, İ. (2025). The impact of online feedback on student learning outcomes: A meta-analysis study. Asia Pacific Education Review.

Evans, C. (2013). Making Sense of Assessment Feedback in Higher Education. Review of Educational Research, 83(1), 70–120.

Kashiha, H. (2025). From algorithms to annotations: Rethinking feedback practices in academic writing through AI-human comparison. Journal of Second Language Writing, 70, 101254.

Li, A. W., & Collins, P. (2026). Formative feedback across sources: Student perceptions and writing outcomes with instructor, peer, and AI-generated feedback. Reading and Writing.

Lopera-Oquendo, C., Lipnevich, A. A., Tomazin, L., Máñez, I., León, S. P., & Beatson, N. (2025). Unpacking Student Responses to Discrepant Peer and Teacher Feedback: A Cross-National Comparison. Contemporary Educational Psychology, 82, 102394.

Sandidge, C., & Schultz, B. F. (2024). Building Connections and Enhancing Learning: Student Perspectives of Traditional Discussion Boards in Online Courses. Journal of Educators Online, 21(4).

Schluer, J., & Brück-Hübner, A. (2025). Diversity of pedagogical feedback designs: Results from a scoping review of feedback research in higher education. Assessment & Evaluation in Higher Education, 50(2), 295–307.

Wang, J., & Fan, W. (2025). The effect of ChatGPT on students’ learning performance, learning perception, and higher-order thinking: Insights from a meta-analysis. Humanities and Social Sciences Communications, 12(1), 621.