Original Research

Transformational feedback practices: A systematic review of perspectives from higher institution educators

Marchant van der Schyff, Doulette Braak

Received: 28 May 2025; Accepted: 24 July 2025; Published: 19 Sept. 2025

Copyright: © 2025. The Author(s). Licensee: AOSIS.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Background: In our competitive world, students tend to place great emphasis on grades in higher education, as a result they often overlook valuable feedback from their educators.

Aim: This study explores innovative approaches, such as ungrading and using artificial intelligence (AI), to enhance student engagement with quality feedback, fostering deeper learning and skill development.

Setting: The study contextualises quality feedback within evolving educational environments influenced by technological advances, particularly generative AI.

Methods: Through a synthesis of peer reviewed articles published on scholarly platforms, such as Google Scholar and EBSCOhost, a systematic review is used to define the study’s purpose, and apply the transformational learning theory to further truncate key perspectives from educators.

Results: The findings of the study suggest that alternatives to traditional grading can enhance student engagement with feedback. These findings provide a foundation for implementing ungrading in higher education as a transformative teaching and learning approach. This approach aims to produce positive, detailed and actionable feedback to enhance student feedback literacy. An objective of this study is to redefine quality feedback considering technologically advanced and socio-economically dynamic educational environment.

Conclusion: Findings suggest that ungrading and generative AI enhance feedback practices in higher education, emphasising the need for ethical policies and considerations regarding AI’s limitations.

Contribution: This study highlight that effective feedback should empower students and learners, support metacognitive development, and promote educational equity, while it suggests that future research explore students’ perceptions of feedback value and the implementation of ungrading to refine feedback processes.

Keywords: quality feedback; generative AI; higher education; ungrading; transformative learning.

Introduction

Students in institutions of higher learning are required to complete assignments, portfolios and write tests and exams to show proficiency in relevant skills to receive their qualifications. The educators of these assessments are tasked with providing students with quality feedback on their work to develop these skills. The problem often is the lack of agreement on what quality feedback is. The effectiveness of providing comments to students for their learning and development by educators who provide scalable, positive and clear comments and not merely grades on assessments cannot be overlooked (Van der Kleij et al. 2015). According to Hough (2023), a ‘grade’ is a way an educator calculates and reports on a student’s performance and is an accumulation of points between 0 and 100. One possible genesis of this practice could be when in 1785, Yale President, Ezra Stiles, awarded grades to his students to identify them either as optimi (excellent), second optimi, inferiors (developing) and pejores (failing). The difference between grading, which is putting a value on an assessment, and providing critiques, which is given to a student so they can learn, is developmental. Grades do not necessarily assist students in moving towards a point of greater understanding, which is what all assessments should do (Hough 2023). Instead, Nieminen and Carless (2023) describe how grading student assessments risks psychologising student-educator feedback behaviour, while promoting feedback literacy carries the potential for reframing processes through the idea of individual skills development. Feedback literacy refers to the capacity of students and educators to optimise the benefits of feedback opportunities (Nieminen & Carless 2023) and is an approach predicated on shared responsibilities between the student and the educators through design, relational and pragmatic dimensions (Carless & Winstone 2023).

Background to the study

In the seminal article, I no longer grade my students’ work – and I wish I had stopped earlier, Elisabeth Gruner (2022) writes that educators rate students’ work by assigning grades with indices of ‘A’ for superior work, while those who failed the assessment receive an ‘F’. Gruner (2022) continues that the practice of grading and ranking students may seem necessary but is highly inequitable, a view supported by Ko (2019), who criticises grades as the result of poorly designed summative assessments that overlook exam psychometrics and affect students’ self-worth. Gruner (2022) explains that approaching student assessment marking in a deconstructed or ungrading way shows promise in understanding the value of quality feedback. For Ko (2019), grades mostly measure how well a student can do tertiary education, but question whether this is a good enough predictor of skill and ability. Although outside of the scope of this study, Ko (2019) evaluates grades by imagining them as an impediment to achieving success in society, masking inequalities such as lack of sleep, poor nutrition, level of interest in a topic, language fluency and ability to manage time properly, which are all necessary to accomplish academic goals. Even academics in defence of the use of grades, such as Nguyen (2022), discuss grades in the context of participation in extra-curricular activities to being more well-rounded by the time students graduate, but do not discuss the benefits of achieving good grades on their own.

This poses three research questions:

• In existing literature, how would educational scholars describe quality feedback?
• According to educational scholars, how do grading practices influence student engagement with quality feedback?
• How is the use of generative artificial intelligence (AI) positioned in scholarly discussions on quality feedback?

Historical practices in crafting assessment feedback

In 2012, the Educational Leadership Conference held in Vancouver published a collective ‘wisdom of authors’ regarding feedback for learning in which they presented seven axioms, including ‘feedback is information about how we are doing’, ‘students use feedback for learning when the lecture room is considered a safe space to make mistakes’ and ‘students need to know what their learning target is, or else comments are just an educator telling them what to do’ (Wiggins et al. 2012). One of the key reflections of the conference was ‘when we give a grade as part of our feedback, students routinely read only as far as the grade’ (Wiggins et al. 2012). This statement supports Ko (2019), who explains how students use grades as a barometer of their skills and a reflection of their ability to learn, which shape their self-efficacy and beliefs about their intelligence and identities. This, in turn, will shape their future success in learning. Why, then, is the practice of assigning grades not scrutinised more? Hough (2023) claims that traditional grading is inconsistent, even among educators of the same module in the same tertiary institution. For example, students may demonstrate mastery of content but might not be able to apply this technical knowledge to real-world scenarios, and for some educators, the former may carry more weight than the latter, causing grade difference. Blum (ed. 2020) claims that historically, grading was ‘standards-based’, meaning that true learning is not subjected to auditing institutional motivations such as corporate reports to investors or the changing mandates from political accreditation bodies. Learning, here, is subordinate to assigning grades to determine proficiency in a subject matter.

Contemporary trends in providing quality feedback

The traditional grading system is not aligned with learning outcomes, while more progressive educational institutions, such as Melrose High School in the United States of America, are assigning mastered or in progress and provide students unlimited chances to submit formative assessments to learn the material and become proficient in a module’s content (Gube 2024). Gube (2024) therefore argues that South African higher education institutions should strategise around the current academic needs of students to support their development in a changing socio-economic climate. Although Gube (2024) does not offer any clear recommendations on what students’ academic needs are, or how they can be met through quality comments, Ntshangase and Msosa (2022) argue that the South African Qualifications Authority (SAQA) makes every effort to generate well-qualified graduates. In their study, Ntshangase and Msosa (2022) reflect on the reputation of SAQA by investigating their policies on, among other things, the feedback provided to students by registered higher education institutions. The authors were concerned with the proliferation of fake degrees in South Africa, and that one of the ways to evidence the legitimacy of a graduate’s qualification is to evidence the assessments they received, including commentary from their educators. Ensuring feedback is immediate or at least timely can improve the quality of feedback and make it more integral part of the student’s developmental and learning process. This approach also involves reducing the overemphasis that tertiary institutions have placed on documenting feedback (Winstone & Boud 2022).

To address the lack of knowledge in South African higher education and the value of the degrees they produce, Van der Kleij et al. (2015) discuss the three most currently used approaches to assessments including data-based decision-making (DBDM), Assessment for Learning (AfL) and diagnostic testing (DT). These approaches to assessing student learning yield some insights to how educators can provide feedback to their students. In the AfL approach, feedback is seen as the most important aspect of a student’s ability to develop skills and to learn. This approach is aligned to Vygotsky’s metacognitivism, which refers to reflecting on one’s learning, and therefore feedback should be given consistently (Van der Kleij et al. 2015).

An example of how Blackboard, a learning management system where students submit their assessments and educators annotate, has features that are engaging but also limited in providing feedback to students:

From the screenshot (Figure 1), there are tools educators can use to provide feedback to students, such as ‘comments’, ‘annotations’ on the actual script and a space to record the grade; however, for the student to benefit from the feedback, they have to (1) consider the educator’s notes beyond the grade, which most students do not do (Hue & Kennedy 2015; Ko 2019; Wiggins et al. 2012), and (2) must be familiar with the educator’s feedback style and have a high level of feedback literacy. In their mixed-methods study, where 50 respondents were interviewed, Simonette and Joseph (2024) confirmed that educators have consensus around the importance of formative assessments, where comments are developmental and provide students with an opportunity to improve their skills, despite the challenges of a more summative-driven educational culture. Most of the educators indicated they found significance for learning in open-ended problem-solving and oral questioning in their response to assessment submissions. To this end, there are different levels of quality critiques, according to Schwartz (2017), which could support educators in engaging their students through their feedback practice.

FIGURE 1: Screenshot of feedback given to an honours student on a formative assessment (assignment) submitted on the Blackboard system.

Figure 2 shows how educators can use current modes and resources in their feedback approach to improve student engagement with their comments and to increase its utility. In Level 1, the educator can substitute comments such as ‘Come and see me!’ with ‘Your work shows a solid grasp of the key concepts, including X and Y’, while Level 2 might extend this comment to include: ‘…however, you need to apply these concepts to an example to improve synthesis and clarity’. In Level 3, the educator can insert a relational comment, for example: ‘Although this is a challenging section, you will be able to master it by referring to resource Z’. While this takes some time for the educator, in summative assessments, the consideration of the student’s abilities should be less intensive if they properly engage with these types of comments. Another way student engagement can be improved, according to Sitto-Kaunda, Moroeng and Makhubela (2023) in a study investigating relatability in formative assessment feedback practices of young Black academics in South Africa, is to connect academic content with meaning making, localised and real-world problems. Therefore, feedback becomes meaningful when it carries a moral responsibility for cognitive and socio-emotional development to build student confidence by including their historical and under-resourced realities.

FIGURE 2: Schwartz’s (2017) framework for assessment as relational practice.

Predictions on how and why assessment feedback will change

Gruner (2022) explains that the practice of pass or fail options instead of grading during the coronavirus disease 2019 (COVID-19) pandemic created a brief disruption for many educators to the practice of giving actual grades. This was done to reduce students’ stress levels because of studying remotely; however, many returned to grading afterwards. These types of interruptions in the 21st century have led to a deeper reflection on the equity of education, especially for ethnic minorities who, through liberation movements and transformation in educational systems, have experienced a marginal growth in access to formal learning (Gruner 2022). However, according to Gube (2024), educators remain ignorant of their considerations of cultural diversity in their pedagogical practices, and assigning a grade is a major contributor to this. Where students from educational privilege tend to achieve higher grades on essays, students who come from resource scarce communities were unable to master content and writing at the same level (Gruner 2022). Providing quality feedback, instead of assigning grades, could address this inequality. Gube (2024) does not provide any recommendations on how this may be achieved; however, Hue and Kennedy (2015) recommend that, through feedback ethos of respect, care and equality, cultural responsiveness can be improved. These findings precede the contemporary reflections from Schwartz (2017) on how quality comments should hold learners ‘accountable without shaming or blaming’, allowing them the opportunity to grow in receptiveness and openness to the comments assigned to their curriculum submissions.

Le and Tran (2024) find that the use of generative AI tools, such as eduAI, in providing instant feedback helped students to evaluate their learning progress and clarified how they could improve on their knowledge step-by-step. The educator could assess the quality of the generated comments in a way that incorporates a nuanced context, such as known personal circumstances of the student, or specific areas where the student could apply their knowledge. To better understand what generative AI is, Popenici and Kerr (2017:2) define it as ‘computing systems that are able to engage in human-like processes such as learning, self-correction, synthesizing, adapting, and use of data for complex processing tasks’. For educators, course developers and education administrators who were concerned about the abilities of generative AI tools to provide quality comments on assessments to students, Le and Tran (2024) published the results from 200 students where eduAI was used. In the study, out of a 5-point scoring system, the participating students indicated that they developed critical, creative and independent thinking skills (4.24), and were able to evaluate their own work against criteria (4.26) by engaging with the feedback on assessment submissions. These scores indicate that the future of quality feedback excludes assigned grades and includes instant and scalable feedback on assessment submissions. Some educators at private institutions of higher education have been using generative AI platforms, such as ChatGPT, to supplement their feedback strategy to provide students with actionable evaluation of their assessment submissions in a fast and highly agreeable manner (Kumar 2023).

As a result of recent technologies and generative AI’s impact on higher education, Hennessey and Mueller (2020) claim that students will have to develop computational thinking, or rather, thinking like a computer scientist, to solve problems and develop relevant competencies and skills in new socio-economic contexts. This need requires design thinking from students, developed by the feedback they receive from educators on their assessments. The literature on how design thinking is being implemented throughout higher education institutions is sparse (Hennessey & Mueller 2020), and not all educational scholars are as eager to embrace generative AI in learning. Kadel et al. (2022) offer some guidance on how educators and curriculum designers can use generative AI in assessment tasks to address some of the concerns on academic integrity, equity of access, feedback and authenticity. One of the key insights to this research are generative AI’s impact on assessment design, evaluation methodologies and feedback literacies. Kadel et al. (2022) write that the integration of AI in education will be ubiquitous, necessitating a continuous reflection to refine the process.

Kadel et al. (2022) recommend an assessment plan for higher educational institutions (HEIs) who experienced a paradigmatic shift because of generative AI, including a five-phase process (see Figure 3). In the first step, assessments were designed according to learning outcomes; thereafter, the involvement of generative AI was detailed to both students and educators. To maintain academic integrity, in step three, the assessed values, such as critical evaluation, analysis, problem-solving and communication skills, were developed into either prompts or rubrics to guide the evaluation or feedback by educators or the generative AI tool. In the final two steps, the assessments were audited for learning or action-based compliance. The requirement of each assessment, which would influence the type of feedback educators provide, would be detailed and the institution could clearly communicate a rationale behind overall assessments at the programme level as well as the subject level (Kadel et al. 2022). Mishra and Varshney (2024) support Kadel et al.’s (2022) findings, adding that enhanced syllabus communication, engaging in ethical dialogues and using varied assessment formats are also required.

FIGURE 3: Proposed assessment design.

While Le and Tran (2024) describe the benefits of AI models in producing feedback that is in-depth and immediate, Kumar (2023) laments that while using commercial AI services to provide comments on student assessments may include pedagogical merits of consistency received instantly, there are questions around its cost, legality and ethics. A key question academics may have to pose is: How will skill confirmation be determined when generative AI causes grading and marking to be obsolete?.

Ungrading as an alternative practice of providing quality feedback

According to Blum (ed. 2020), learners often strive to achieve good grades instead of using other metrics to measure their progress, such as practicing skills, reflecting on growth before and after a course or the application of practical solutions to real-world problems. Students have been conditioned to measure success through a grade allocation, and not by reflecting on their skills or developmental progress. This idea is fully entrenched in pupils by formal schooling systems usually by the age of 11 already and is used as currency by HEIs (Gruner 2022).

The term ungrading was coined by educational researcher, Jesse Stommel (Morris 2022). The idea of ungrading as an alternative approach to the historical marking of assessments and grading of student articles was popularised through an article by Elisabeth Gruner (2022), an English Professor at the University of Richmond who reflects on her teaching and learning pedagogy during the time of COVID-19. In the article, Gruner (2022) explores three reasons for pivoting from the use of grades. There needed to be a more concerted effort for students to engage with the feedback given by educators – when a grade appears on the script, students focus solely on that. Gruner (2022) claims that when she grades a student’s article, she is grading their background, which is a challenge to an inclusive pedagogy as well as equity. Gruner (2022) admits that she hates grading but loves teaching, and feedback serves the latter but is constrained by the former. Ungrading is an approach to marking assessments that moves away from rubrics and an overreliance on grades to narrative feedback, which provides students with agency to intervene in their own education (Morris 2022). Morris (2022) reflects on how the COVID-19 pandemic resulted in a rare phenomenon where educators and students agreed that contemporary grading practices and measuring of competencies were incongruous, and ungrading, as a relational practice, required the students and educators to engage with one another in a feedback loop.

Making sense of the collected data through the transformational learning theory

Transformative learning involves experiencing a deep, structural shift in the basic premises of feelings, actions and thoughts. It is a shift of consciousness that dramatically and irreversibly alters our way of being in the world, when a learner responds to a disorienting dilemma, questioning and revising their perspectives, engaging in discourses and acting on a new perspective (Cranton 2016). Two contributing transformational learning theorists, Jack Mezirow (1994) and John Dirkx (1998), were referenced to provide a framework that collected and collated data. Mezirow (1994) highlights how moving away from the status quo, such as traditional grading towards formative assessments and ungrading, would encourage critical self-reflection among students and educators. This transformation allows students to focus on their learning process rather than merely striving for grades, promoting deeper learning and critical thinking, while Dirkx (1998) emphasises the role of emotions, such as positive engagement with feedback and the subconscious in transformative learning. Dirkx (1998) would find the ungrading model beneficial in its ability to create an emotionally supportive and engaging educational environment. The shift from grades to detailed feedback and dialogue addresses the emotional and psychological aspects of learning. This method could encourage educators and students to engage more deeply and authentically, reducing stress and creating a more passionate teaching and learning environment (Carless & Winstone 2023; Gruner 2022; Nieminen & Carless 2023). Dirkx (1998) focuses on the affective emotive domain, which complements the idea of only presenting formative assessments (Ko 2019), fostering a holistic educational experience, where emotional engagement is crucial for deep learning (Hennessey & Mueller 2020).

Research methods and design

The research took on a relativist ontology to gain a deeper understanding of higher education educators’ perspectives on qualitative feedback, which would foreground the complexity and uniqueness of assessment submissions instead of trying to generalise on the basis of quantifiable outcomes for the entire population (Pham 2018). By initially identifying 65 academic articles over a 3-month period using purposive Boolean online searches on Google Scholar, Mendeley and EBSCOhost, this study collected data relevant to the research objective, according to the transformational learning theory’s interpretations by Mezirow (1994) and Dirkx (1998). The inclusion criteria required articles to focus on themes of ‘feedback literacy’, ‘quality feedback’, ‘ungrading’, ‘assessment practices’ and ‘higher education’. The articles needed to appear in peer-reviewed journals listed in DOAJ, DHET or Scopus, and the articles needed to be published between 2015 and 2024. As recommended by Largan and Morris (2019), a Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) diagram was used to document the selection process, which included the number of identified articles, exclusion- and inclusion criteria, to ensure replicability and enhance transparency. Following this screening process, 15 articles, which were from studies conducted locally and internationally, were purposively sampled for an in-depth analysis focused on educators’ experiences with feedback practices. Approximately 23% (15/65) of the identified population was sampled because repetitive themes started to emerge, and it could therefore be argued that theoretical saturation was reached. Also, the smaller sample allowed for greater reflexivity and analytical rigour, as the purpose of this study was to provide a systematic, in-depth review from a diverse range of articles within its practical constraints. The process was supported by consultation with information specialists at a higher education institution.

In the absence of speaking to all educators and researchers on quality feedback, this study includes collected insights and findings from primary and secondary sources. While informal primary discussions with an educator were included, the main data were extrapolated from published articles in relevant journals as secondary sources. Researcher control and scope were considered in this qualitative approach to better understand the context of the themes to address the research problem. This is because the sources were peer reviewed and published for general consumption by the wider academic community and can therefore be used for descriptive aims. This research collates the common findings of relevant articles to develop themes referring to the issues experienced with assigning grades to assessments and ungrading as an alternative, what it means to provide quality comments by educators and the significance of feedback literacy for both the learner and the educator. The articles were collected cross-sectionally to reflect on what is written on the state of assessment comments for students in the current socio-economic and technological environment. The challenge, in qualitative studies, is to ensure that the vast amount of information is evaluated to serve the research problem and to critique the relevant sections of the articles to ensure the research quality (Bassot 2022; Largan & Morris 2019). Therefore, the identification and selection of articles were important.

Systematic review process

A systematic review was conducted on existing literature describing critiquing practices in higher education. The systematic review process was initially developed for medical sciences but has gained popularity in social sciences, including education, over the years (Chapman 2021; Snyder 2019). A systematic review, known as a secondary data research study, is defined by Kitchenham (2014) as the process where relevant literature, such as published articles, is identified, critically analysed and interpreted to answer specific research questions. This definition is corroborated by Chapman (2021) in questions or topics of interest. The first step in a systematic review is that the research questions must be formulated. The three questions posed were: ‘In existing literature, how would educational scholars describe quality feedback?’, ‘According to educational scholars, how does grading practices influence student engagement with quality feedback?’ and ‘How is the use of generative AI positioned in scholarly discussions on quality feedback?’ Each article was read in full, and a thematic analysis was used to identify recurring insights. From emergent patterns, attention was paid to ‘the definitions and significance of quality feedback practices’, ‘educators’ challenges with assigning grades’, ‘ungrading as an alternative assessment practice’, ‘the significance of feedback literacy to students and educators’ and ‘the integration of AI tools in feedback processes’. Each study was evaluated for its relevance to the research questions and its contribution to contemporary discourses (Zawacki-Richter et al. 2020). Therefore, the themes are evidence-based to provide insights for future research and practices in higher education institution feedback mechanisms. Because of the qualitative nature of the research, it was not necessary to identify all research articles related to feedback practices in higher education as conceptual saturation was more important (Thomas & Harden 2008).

While systematic reviews traditionally aim for comprehension, in qualitative investigations such as this study, conceptual saturation, rather than exhaustive coverage, is prioritised (Thomas & Harden 2008). Therefore, the review focused on achieving deeper understanding through the validity of themes rather than including all 65 articles on the topic. A summary of the literature is provided in Table 1.

The thematic analysis was performed in three stages, with the first two stages involving the interrelated steps of coding the text and then the development of descriptive themes. Each article was read to identify the relevant segments of text and assigned to open codes. codes were inductive; however, they were also informed by the research questions, and the organisation of the codes was completed accordingly and grouped into thematic clusters where recurring themes relate to quality feedback, grading practices and ungrading, and the use of AI was identified. Mezirow (1994) and Dirkx’s (1998) principles of transformational learning were used as a theoretical lens to synthesise and interpret these themes. The authors shared a codebook and independently coded a subset of 50% of the sample articles, while the final coding was jointly reviewed to ensure inter-rater reliability across the data set. Initial codes included quality feedback practices, integration of generative AI in feedback, theoretical connections to transformational learning and grading practices such as ungrading in student engagement.

Discussion of findings

Alternatives to assigning grades to (summative) assessments, such as brief semi-structured interviews – those used by Harvard University – that can be used to assess whether students have developed the skills required to pass a module, or exclusively focussing on formative assessments developed with a growth mindset and engaging students through providing quality feedback (Ko 2019). Generative AI can also be used here to process data in a complex and targeted way for students to use in self-correcting (Popenici & Kerr 2017).

Rethinking grading and moving towards an ungrading model might keep educators in the profession longer (Hough 2023). There is a crisis in the retention of qualified educators because they become more excited about teaching than assigning grades to meet institutional success rates. Gruner (2022) states, ‘Feeling freed from the tyranny of determining a grade, I wrote meaningful comments, suggested improvements, asked questions and entered into a dialogue with my students that felt more productive’, and created enthusiasm for teaching. The problem with marking assessments, according to Hue and Kennedy (2015), is that grades are a challenge to the cultural responsiveness of educators as they fail to integrate the ‘part’ of assessment with the ‘whole’ of teaching. Marking therefore fails to manage diverse learning needs of students, and language proficiency becomes a determining factor for academic success, while generative AI can assist in synthesis by engaging students with quality feedback and their learning goals (Popenici & Kerr 2017), and not pure grading. Hue and Kennedy (2015) also criticise grades as highly inequitable. In their study, Nieminen and Carless (2023) further support these criticisms by analysing 49 published articles on feedback literacy. They explore how grade-motivated feedback as external input is opposed to feedback as a psychological developmental process residing within students, concluding that the latter carried positive implications for student-educator empowerment. Winstone and Boud (2022) echo Hue and Kennedy’s (2015) concern that students focus too much on grades and add that feedback has become subordinate to other considerations in course design. Blum (ed. 2020) claims that ranking students is counterproductive for learning and is immoral. Institutions of higher learning that prioritise distinction rates above learning competencies create students less interested in learning who prefer easier tasks and shallower thinking to achieve higher grades. Blum (ed. 2020) recommends doing away with dispensing ratings, such as grades, and favouring narrative reports that provide quality and qualitative feedback. This approach aims to redefine what we call ‘feedback’, which is often merely seen as a defence of the grade awarded. Grades impeded students’ ability to engage with comments to stimulate agency and skills development (Schwartz 2017).

Educational scholars describe quality feedback in several ways. An example of quality feedback came from a study by Makrides and Yeates (2022) on video-based feedback (VbF) and found that videos enhanced students’ memories of their performance in presentation-based assessments. This enhancement in memory made students more receptive to educators’ feedback and increased the perceived credibility of the feedback provided. Reflecting on video performances produced insights for students beyond the grades assigned by educators. Students triangulated these novel insights with their own self-assessment and experiences from practice to reflect deeply on their performance, which led to achieving additional learning objectives. Although the study by Makrides and Yeates (2022) suggests a more practical redevelopment of assessments than the processes describing feedback, it did provide insights on the benefit of quality feedback.

Some educators and parents are concerned that moving away from a traditional form of letter grading systems would result in students not receiving adequate feedback on their work (Schneider 2020). However, as a method of providing quality criticism on assessments, educators would give students extensive feedback and multiple opportunities to revise their work. At the end of the academic semester, the student would submit a portfolio of revised work, along with an essay where they evaluate their learning. Although a grade for each module is required and the student would include a grade in their reflective essay, the educator reserves the right to change the grade. There are three reasons for this model, including wanting students to focus on the feedback and not stop when reading the grade, wanting to move away from grading a student’s background, where students with educational privilege generally achieved higher grades, and to become enthusiastic about teaching again, because grading can be quite tedious, while providing feedback on assessments feels more like entering a dialogue with a student (Gruner 2022).

The role of generative AI in the commenting process has emerged as a transformative tool, offering significant advantages in terms of speed, consistency and detail. Recent research highlights its potential benefits and limitations, as well as the need for AI literacy and clear policies to maximise its effectiveness in educational settings (Kadel et al. 2022). Mishra and Varshney (2024) emphasise the importance of generative AI literacy in education, arguing that understanding how to use these tools can significantly enhance their educational value. Proficiency in AI tools allows both students and educators to effectively craft answers and provide feedback, making the feedback process more efficient and meaningful. Generative AI tools, such as ChatGPT, have shown notable efficiency in grading assessments. A pilot study by Van der Schyff (2024) demonstrates that ChatGPT 3.0, when prompted with a rubric, graded a postgraduate student’s formative assessment with an 83% score, while the educator awarded 78%. This 5% variance indicated that AI could grade a student’s assessment submission with reasonable accuracy. Additionally, ChatGPT 3.0 provided detailed feedback within minutes, highlighting its capability for rapid delivery (Le & Tran 2024). This is supported by Wiggins et al. (2012), who emphasise that feedback is most effective when students have adequate time to act on it. However, the AI’s feedback often included vague generalities and tautological comments, requiring the educator to review and refine these to ensure their accuracy and usefulness. Kumar (2023) adds that AI-generated feedback tends to be discrete, convenient and consistent, which can enhance fairness in the assessment process. However, Kumar (2023) also raises concerns about academic integrity, warning that reliance on AI tools must be balanced with a rigorous understanding of their limitations and potential for misuse.

Kadel et al. (2022) stress the necessity for clear policies governing the use of generative AI in educational assessments and feedback. They propose a structured five-step approach to integrate AI tools effectively:

1. Implement training sessions to develop AI literacy among students and educators.

2. Ensure continuous monitoring and improvement of AI tools.

3. Establish clear guidelines on the ethical use of AI in assessments.

4. Develop mechanisms to validate and verify the accuracy of AI-generated feedback.

5. Encourage active involvement of educators in the feedback process to supplement AI-generated responses with expert insights.

There is an understandable apprehension among educators regarding the ethical implications of using generative AI. According to the United Nations Educational, Scientific and Cultural Organization’s (UNESCO) AI Ethics Framework, some recommendations to address these concerns include a call for hosting and development companies of generative AI tools to develop test protocols to ensure the fundamental human rights of users, as well as to protect them from cyber-attacks and to provide equitable and unbiased access. It also calls for educational AI to align with global goals, foster public awareness and promote AI literacy (Rakha 2023). Furthermore, the European Union’s (EU’s) Guidelines for Trustworthy AI propose requirements for AI in education to mitigate bias and improve accountability and transparency. These include a technically robust and secure firewall, a design that respects users’ privacy, and is human-centric and inclusive (European Commission 2019).

Key insights and implications of findings

Educators’ key insights about feedback

According to Simonette and Joseph (2024), educators emphasise the importance of formative assessments but highlight significant challenges, such as the time required to create and evaluate these and the varying skill levels of academics in adapting to new educational technologies. Eliminating summative assessments is recommended to create more resources, such as time and energy, to improve quality feedback given on formative assessments (Ko 2019). This shift would allow educators to focus more on providing knowledge-building feedback within formative assessments. These insights suggest a need for professional development and structural changes in assessment strategies. Here, educators could transition from traditional grading to considering alternatives by actively including feedback literacy in their professional development strategy. They can accomplish this by attending workshops on constructive feedback and developing formative, narrative and practical pedagogies, engaging in peer review and community of practice initiatives, embedding reflections on feedback in their portfolio of evidence and joining mentorship programmes with a focus on how to ethically use generative AI tools. Blum (ed. 2020) contributes to this idea by stating that narrative reports with built-in feedback protocols should be provided instead of assigning grades.

Importance of feedback literacy

Developing feedback literacy is essential for students to engage with assessment feedback in a meaningful way, promoting their growth and learning (Carless & Winstone 2023; Nieminen & Carless 2023). This focus ensures feedback is both ethically sound and pedagogically effective, enhancing the educational experience and fostering student development. This will be a critical skill for students to develop beyond their academic training into the world of work, where incorporating feedback will determine their success in rapidly changing technological and socio-economic contexts. In transformational learning theory, the use of generative AI in feedback should be a tool for enhancing critical reflection and using assessments as a relational practice and must be used ethically to complement human feedback and not replace it (Schwartz 2017). Artificial intelligence can offer quick, detailed responses that help students reflect on their performance; however, educators should caution against over-reliance on AI, emphasising the need for emotional and human elements in providing quality feedback (Carless & Winstone 2023). This dedication to feedback literacy forces educators and students to engage in deeper metacognitive processes that can lead to transformative shifts in their learning. Although not the scope of this study, Manzi and Moreeng (2023) extend the benefits of feedback literacy among educators as well by stating that even though there are case studies of effective feedback practices in South African HEIs, overall inconsistencies highlight the need for training and capacity building to develop meaningful feedback practices.

Why grading is problematic – Ungrading as an alternative

Educational theorists consistently highlight traditional grading systems as problematic because of its inherent inequities, demotivating effects and the stress it induces in students (Ko 2019). As an alternative, ungrading offers a promising approach. Instead of assigning numerical or letter grades, assessments are marked as in progress, mastered or mastered with distinction, allowing students unlimited opportunities to improve (Hough 2023). This shift towards a competency-based model emphasises ongoing learning and skill development. Furthermore, ungrading could enhance educational equity, particularly benefiting ethnic minorities by using feedback as a tool for continuous improvement and support (Gube 2024). By focusing on mastery and personal growth, ungrading could reduce stress and foster a more inclusive and supportive learning environment. The ungrading approach, where students focus on quality narrative feedback and self-evaluation, aligns with Mezirow’s (1994) focus on critical reflection and Dirkx’s (1998) emphasis on emotional involvement in learning. This method fosters a more supportive and dialogic educational environment, promoting transformative learning experiences. Ungrading could be aligned with the Department of Higher Education’s focus on equity and inclusive learning through institutional quality enhancement plans (QEPs) (Department of Higher Education and Training 2020) by training educators usings models such as AfL and evaluating pilot outcomes and demonstrating its impact on both educators and learners. From the literature consulted, there is limited information on ungrading and the practical and relational applications thereof. Schneider (2020), however, cautions that doing away with letter grading systems completely, which seems unlikely in the near future, could negatively impact student motivation, which was the original aim for assigning grades. Educators in the latter half of the 19th century found that students tended to work harder if there was a goal for them to reach (Coleman 2023). To address apathy, procrastination and disengagement of students with quality feedback that does not include a grade-component, the transformational learning theory provides a lens emphasising affective engagement reflected in the use of video-based and narrative comments, which enhance personal meaning making and evoke emotional responses, thereby supporting holistic learning.

What feedback should accomplish

Effective comments are a cornerstone of the educational process, and the research suggests a change in basic assumptions towards enhancing student agency and active engagement. Moni (2021) advocates for feedback that empowers students, enabling them to take ownership of their learning journey. This shift towards student agency places students at the centre of the feedback process, encouraging self-regulation and independent learning. Feedback should include a reflective component, where they provide interpretations to educators about what remains unclear. According to Wiggins et al. (2012), such student feedback can be more impactful than traditional educator-to-student feedback, as it highlights specific areas where students need further clarification and support. The feedback can be provided according to the level (Schwartz 2017) and need. Mezirow (1994) and Dirkx (1998) underscore the benefits of moving away from traditional grading and emphasise the engagement of students in formative assessments (Moni 2021; Schwartz 2017; Wiggins et al. 2012). One of the ways this can be accomplished is for educators to consider the student’s context so that the feedback may empower them (Sitto-Kaunda et al. 2023).

Introducing metacognitivism into the feedback process further enhances its effectiveness. Van der Kleij et al. (2015) explain that metacognitive feedback supports immediacy, creates a continuous feedback loop and focuses on how students learn. This approach ensures feedback is regularly integrated into the learning process, creating a praxis that helps students develop critical thinking and self-awareness. Another approach, AfL, emphasises the quality of the learning process and the provision of high-quality comments (Van der Kleij et al. 2015). In this model, criticism is not an end, but a means to improve the overall learning experience. Quality feedback is practical, clear and engaging, as emphasised by Winstone and Boud (2022). They argue that for comments to be developmental, students must be active participants, working with and applying the information provided by educators. This engagement transforms feedback from a passive receipt of information to an active, collaborative process. Moreover, quality feedback is characterised by respect, care and equality, which enhances multiculturalism in education (Hue & Kennedy 2015). Respectful and caring comments acknowledge the diverse backgrounds and experiences of students, promoting an inclusive educational environment. This approach ensures that all students, regardless of their cultural or ethnic background, receive feedback that is equitable and supportive. Therefore, ungrading, as an alternative method of providing quality feedback to students, fosters an educational environment that promotes critical reflection (Hue & Kennedy 2015), emotional engagement and deeper learning (Dirkx 1998; Winstone & Boud 2022), aligning with the core principles of transformational learning. Two major benefits that grading has over other models of feedback, such as ungrading, are that it helps to keep the student accountable through a universally understood metric. Everyone knows what grades mean and a precise scale of performance can help a student to build self-analytical skills (Coleman 2023). However, this assumes that all educators can standardise grading and accurately quantify the strengths and weaknesses of their student assessments.

Lastly, feedback should address the competency levels of students, as this can legitimise their qualifications (Ntshangase & Msosa 2022). Students can be classified as ‘mastered with distinction’, ‘mastered’ or ‘still in progress’ to indicate their level of achievement and identify areas where improvement is needed. Competency-based feedback provides clear benchmarks for students to understand their progress and areas for improvement. This clarity helps them to build the necessary skills and knowledge required for their academic and professional success. The findings reveal a systemic shift from traditional grading towards more formative, feedback-driven learning. Educators stress the value of meaningful feedback but face challenges like time constraints and uneven digital literacy, which is central to student development (Simonette & Joseph 2024). Moving away from summative assessments could free up resources for more impactful, knowledge-building feedback (Ko 2019). Feedback literacy emerges as essential for fostering student agency, critical reflection and lifelong learning (Carless & Winstone 2023). When grounded in relational and inclusive practices, feedback becomes a tool for equity and transformation (Hue & Kennedy 2015; Mezirow 1994). Educational reform is required to include alternatives like ungrading and competency-based models to reduce stress, promote emotional engagement and reflect students’ growth better (Gruner 2022; Gube 2024). Policymakers, curriculum designers and educators are encouraged to draw on the findings of this research for future studies, including empirical research involving students.

Conclusion

In conclusion, this study focused on feedback practices in higher education. The results suggest that ungrading as a practice can be beneficial for student development and that generative AI can be used as a tool for quality feedback. When used effectively, generative AI can provide educators with guidance when supporting Mezirow’s (1994) transformational learning theory of timely feedback that prompts critical self-reflection, while also complementing Dirkx’s (1998) view that quality feedback must be emotionally resonant and include personalised learning interactions. However, its limitations must be considered, and policies must be put in place for its ethical use as a tool for feedback and marking.

Because of this study focusing on existing literature, it is limited by its lack of empirical insights from HEI student; therefore, future research should investigate students’ perception of the value of feedback in creating a learning loop and praxis, and the merit they assign to feedback compared to grades. A further study into the practical implementation of ungrading models across disciplines and institutions, with attention to educator workload, student motivation and policy constraints, can further explore feedback processes. Drawing from Kadel et al. (2022), HEIs can practically transition from traditional grading practices to ungrading by designing assessments that focus on learning outcomes, while clearly communicating assessment expectations and purposes. These assessments could be redesigned to align with module outcomes using competency-based language, for example, ‘in progress’, ‘mastered’ or ‘mastered with distinction’, and replacing grades with narrative feedback rubrics. Feedback literacy, here, could be facilitated and improved through course orientation workshops (Mishra & Varshney 2024). Another practical recommendation could be to establish ungrading pilot studies grounded in transformational learning principles and student reflexivity (Gruner 2022). The pilot programme could be introduced in selected first-year modules, where HEI students are asked to keep journals and provide video testimonials on their understanding of feedback and how this creates meaning for them. Although this study provides a better understanding of current grading and feedback practices in HEI, future longitudinal studies can enhance the themes of ethical and pedagogical role of generative AI in formative assessments to address concerns related to bias, over-reliance and providing relational support, and strategies for including feedback literacy into curricula.

It is clear that, for quality feedback to be effective, it should accomplish several key objectives: empowering students, including reflective components, supporting metacognitive development and emphasising quality and clarity. Feedback should also foster multicultural respect and equality, and address competency levels. By achieving these goals, feedback becomes a powerful tool for enhancing learning, promoting student engagement and ensuring educational equity.

Acknowledgements

The authors declare that they have no financial or personal relationships that may have inappropriately influenced them in writing this article.

M.v.d.S. has conceptualised, performed the research, written the first draft, edited and revised the research, while D.B. collaborated at each action point as co-author.

Ethical clearance to conduct this study was obtained from The Independent Institute of Education Varsity College Ethics Committee (reference no: R.000164 [REC]) in 2023.

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

The authors confirm that the data supporting the findings of this study are available within the article and its references.

The views and opinions expressed in this article are those of the authors and are the product of professional research. They do not necessarily reflect the official policy or position of any affiliated institution, funder, agency, or that of the publisher. The authors are responsible for this article’s results, findings, and content.

References