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 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.

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 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.

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.

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?.

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.

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).

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.