I read foundational papers for the same reason I read newer ones, to test what still works when the conditions around the paper change. Carless and Boud’s 2018 piece on student feedback literacy is one of those papers. It was written before generative AI showed up in classrooms, and it builds an argument about feedback that has only gotten more relevant since.
The premise is simple. Feedback only does its work when students know how to use it. Without that, comments accumulate on assignments and learning never actually shifts. Carless and Boud (2018) argue the field had spent too much time on what teachers say and not enough on what students do with what teachers say. They define student feedback literacy as the understandings, capacities, and dispositions students need to make sense of feedback information and use it to improve their work or learning strategies.
That definition does real work. It moves feedback away from a one-way transmission and toward a relationship students have to learn to participate in. Carless and Boud (2018) build their definition on Sutton’s earlier academic-literacies work, then add two layers Sutton didn’t address: emotion and action.
The Four Features of Student Feedback Literacy
Carless and Boud (2018) propose a framework with four interrelated features: valuing feedback information, making judgments, managing affect, and taking action. The first three feed into the fourth, which is where actual learning improvement shows up.
The first feature covers students recognising the value of feedback, understanding their active role in the process, and grasping that useful comments arrive from many sources beyond the teacher. The authors fold technology into this dimension as a way for students to access, store, and revisit feedback over the course of a programme. They describe it as the foundation everything else rests on.
The second feature is making judgments, which the authors define as the capacity for evaluative judgment. Students need to be able to assess the quality of their own work and the work of peers. Carless and Boud (2018) tie this directly to peer feedback as the productive vehicle for building that capacity.
Managing affect comes third in the framework. This covers the emotional side of receiving criticism, particularly when comments touch grades or identity. Students often react defensively under those conditions, and the authors name affect as a real barrier to engagement that learners need to learn to handle.
The fourth feature is taking action, which the authors say is the most often underdeveloped of the four. Students need motivation, a clear opportunity, and the strategies to actually use the comments they receive. Without action, the first three features never produce learning improvement.
Their argument crystallises in one of the lines I think about most when I read about feedback now. Carless and Boud (2018) contend that “information becomes feedback only when students act on it to improve work or learning strategies” (p. 1322). That sentence reframes everything. A comment on a draft is information. It only becomes feedback if the student uses it.
Why This Framework Aged Well
I read this paper now because I notice the AI feedback conversation skipping past the student altogether. Tools generate comments. The next round of studies measures whether those comments are accurate. Vendors race to bring it all to scale. Almost no one asks the prior question Carless and Boud asked in 2018: does the student have the literacy to make use of the comments at all?
Recent work by Hawkins, Taylor-Griffiths and Lodge (2025) on feedback literacy and AI-enhanced essay writing extends this exact concern into the GenAI era. Students who lack feedback literacy don’t suddenly gain it because the comments are coming from a chatbot. If anything, AI-generated feedback amplifies the risk, because the comments sound authoritative, arrive instantly, and feel low-stakes to ignore.
The two enabling activities Carless and Boud (2018) suggest, peer feedback and analysing exemplars, also age well. Their argument that composing comments for peers demands cognitive work that receiving feedback doesn’t require holds up cleanly in the era of AI tutors. Peer feedback puts students in the judgement seat. AI tutors put them right back in the recipient seat, where the burden of evaluation gets handed off to the system. That’s not the same cognitive work.
The authors do warn that peer feedback only delivers gains when students are coached in giving and receiving it. Carless and Boud (2018) argue that “without training and support for peer feedback, anticipated gains are unlikely to occur” (p. 1320). I’d argue that the same warning now applies to AI-generated feedback, with even higher stakes. We’re handing students a feedback source they’ve never been taught to evaluate, and assuming the receiving end will sort itself out.
Where the Framework Needs an Update
The 2018 paper assumes feedback comes from teachers and peers. That assumption is no longer accurate. In a 2026 classroom, feedback information also arrives from AI tools, automated revision systems, and generative chatbots students consult on their own time. Carless and Boud’s category of “different sources” needs an explicit AI subcategory now, with all the literacy demands that brings.
The first feature was originally about students seeing the value of comments from teachers and peers. The 2026 version has to add a layer about students learning to evaluate the source itself. Is this AI tool calibrated for my discipline? Is it confidently wrong? Does it match what my teacher would say? Those are literacy questions Carless and Boud could not have anticipated, but their framework handles them well once you build the AI source in.
Sperber et al.’s (2025) work on PAIRR formative assessment offers one practical extension, using AI inside a peer review process to coach students through better commentary. That kind of design respects the original Carless and Boud argument and takes the AI source seriously at the same time.
What Teachers Have to Do
Carless and Boud (2018) place teachers in a facilitating role, not a content-delivery role. Teachers design the curriculum so feedback can be used, hold meta-dialogues with students about feedback processes, and coach students through giving and receiving comments. The authors propose that “through repeated experiences of making self-evaluations, students learn to generate internal feedback and gradually acquire expertise in making more sophisticated academic judgments” (p. 1322).
That’s a curriculum design argument, not a single-assignment fix. Iterative tasks, multi-stage projects, and e-portfolios are the structural conditions feedback literacy needs. One-off assignments with terminal grades teach students that feedback is just a justification for a mark.
This is also where Fan et al.’s (2025) findings on metacognitive laziness connect. Students with weak self-regulation outsource the thinking to AI without engaging the comments. The Carless and Boud framework predicts that outcome, because those students never developed the action dimension. The gap was already there. The technology exposed it.
The Lifelong Learning Argument
The paper closes on a point that has only gotten stronger. Carless and Boud (2018) argue that “feedback literacy is not just a tool for doing better in university studies but a core capability for the workplace and lifelong learning” (p. 1323). In 2026, with AI assistants offering feedback inside every workplace tool, that argument is no longer aspirational. It’s operational. Knowledge workers receive AI-generated commentary on their writing, code, and decisions every day. Whether they treat that commentary as useful information or as authoritative judgment depends on the literacy they built years earlier.
A 2018 framework can’t cover all of what 2026 needs. But this one named the right four problems, in the right order, and gave teachers a structure to build the work around. Read it again with AI sources in mind, and it reads like a paper written ahead of its moment.
References
- Carless, D., & Boud, D. (2018). The development of student feedback literacy: enabling uptake of feedback. Assessment & Evaluation in Higher Education, 43(8), 1315-1325. https://doi.org/10.1080/02602938.2018.1463354
- Fan, Y., Tang, L., Le, H., Shen, K., Tan, S., Zhao, Y., Shen, Y., Li, X., & Gašević, D. (2025). Beware of metacognitive laziness: Effects of generative artificial intelligence on learning motivation, processes, and performance. British Journal of Educational Technology, 56(2), 489–530. https://doi.org/10.1111/bjet.13544
- Hawkins, B., Taylor-Griffiths, D., & Lodge, J. M. (2025). Summarise, elaborate, try again: Exploring the effect of feedback literacy on AI-enhanced essay writing. Assessment & Evaluation in Higher Education. https://doi.org/10.1080/02602938.2025.2492070
- Sperber, L., MacArthur, M., Minnillo, S., Stillman, N., & Whithaus, C. (2025). Peer and AI Review + Reflection (PAIRR): A human-centered approach to formative assessment. Computers and Composition, 76, 102921. https://doi.org/10.1016/j.compcom.2025.102921