The Challenge
For lecturers at technical universities across Ghana, grading is not just time-consuming, it is structurally broken. Class sizes routinely exceed 100 students per cohort, and a single lecturer is often responsible for multiple classes running concurrently.
A lecturer from one of these universities reached out to see how GradePoint AI could help. At his institution, lecturers are allowed up to six weeks to grade assignments and submit results. In practice, that deadline is rarely met. Grading typically stretches to the end of the semester, meaning students never see their feedback while it could still make a difference.
The lecturer teaches Integrated Marketing Communication across two classes (an Evening Session of 122 students and a Weekend cohort of 91 students) totaling 213 submissions for the same assignment.
The university has no access to plagiarism detection software or AI screening tools, leaving it entirely to the lecturer to spot patterns of copying or AI-generated content. When grading hundreds of submissions manually, the law of diminishing returns sets in quickly. Integrity checking becomes near-impossible as fatigue accumulates. Suspected violations go uninvestigated. Standards erode quietly.
The stakes were also unusually high. A mid-semester exam was scheduled within one week of the submission deadline. For feedback to have any meaningful impact on student preparation, it needed to be returned within days not weeks, and certainly not at semester's end.
The Assignment
Students were tasked with designing a comprehensive Integrated Marketing Communication plan for a mutual fund seeking to grow its customer base by 1,000 subscribers. The assignment required students to apply a hierarchy of effects model, develop SMART objectives, propose multi-channel IMC strategies, design a 12-month media schedule within a GH₵150,000 budget, and establish monitoring mechanisms.
It was a demanding, real-world brief that required both theoretical command and practical financial thinking. This was exactly the kind of assignment that reveals the difference between genuine understanding and surface-level content generation.
The GradePoint AI Solution
The lecturer partnered with GradePoint AI to process both classes simultaneously, using his existing marking scheme and rubric. The solution covered everything from personalised student feedback to class-level analytics and academic integrity screening – all within a single workflow.
Custom Rubric Integration
- • The lecturer's marking scheme was embedded directly into the grading workflow
- • Ensuring consistent, criteria-aligned evaluation across all 213 submissions
AI Detection Screening
- • Every submission was analysed for AI-generated content
- • Those exceeding a 30% threshold automatically flagged for review
Plagiarism Cross-Comparison
- • Submissions were compared against one another
- • Surface copying and collusion across both cohorts
Personalised Feedback at Scale
- • Each student received individual written feedback
- • Identifying specific strengths, gaps, and targeted recommendations
Cumulative Evaluation Reports
- • A class-level analytics report was produced for each cohort
- • Covering grade distributions, common improvement areas, top performers, at-risk students, and academic integrity flags
Lecturer Approval Workflow
- • No feedback reached a student without the lecturer's review and explicit approval
The Results
Efficiency Gains
Both classes were fully processed and feedback-ready within three days – a turnaround the lecturer described as unprecedented in his career. Under the manual grading approach, this volume would have taken weeks, with results unlikely to reach students before the end of semester.
Student Impact
With feedback returned before the mid-semester exam, students had a rare and meaningful opportunity: the chance to understand where they went wrong and apply those lessons immediately. For the first time, assignment feedback functioned as a teaching tool rather than a retrospective grade slip.
Both classes showed similar areas of weakness that included surface-level application of the AIDA model, failure to adhere to the GH₵150,000 budget constraint, and a tendency to list tactics rather than develop genuinely integrated strategies. With GradePoint AI's cumulative insights, the lecturer was able to address these gaps directly in class ahead of the exam.
Academic Integrity Insights
The academic integrity findings were significant. Across both classes, 134 out of 213 students (63% of the combined cohort) were flagged for either confirmed plagiarism or AI detection scores of 30% or higher.
For an institution with no dedicated detection infrastructure, these findings would have been entirely invisible under a manual grading approach. Their surfacing now carries a secondary benefit: students are now aware that the lecturer has the capability to detect both AI use and copying – a deterrent effect that is likely to shape behaviour in future assessments.
The Lecturer's Perspective
"GradePoint AI is a very good tool. It makes the work much easier than what we are currently doing. What I appreciated most is that it's not just assessing the students; it's also telling the lecturer where to focus if you want to improve. It gives the strengths and improvement areas to the students, while identifying focus areas for me as a lecturer that I can use to enhance the course content going forward.
When the academic integrity report came in, I put the whole report on the class platform so all the students could see it. This was the first time they had seen a real-life example of how AI can expose them. They are now aware. If I have to teach those students again, next time they will be more responsible, and they will use AI as a learning tool and not to do all the work for them.
What this tool also does is increase the frequency of student assessment. I can now do three assessments in a semester, or even more. Currently, some lecturers may not even do assessments, or they do it but they will not mark it. If this becomes an institutionalised policy, then no lecturer would have an excuse not to assess their students. All you need is to set the question. The students do it, they upload it, and then the assessment is done."
— Dr. Divine Akwensivie, Accra Technical University, Ghana
Key Insights
Feedback Timing Is a Teaching Decision
When feedback arrives at the end of a semester, it functions as an administrative record. When it arrives within days of submission, it becomes instruction. The difference in student outcomes between these two scenarios is not marginal, it is fundamental.
Academic Integrity Is Invisible Without Tools
With no institutional access to detection software, integrity violations at scale go entirely undetected. A 63% flag rate across two classes suggests this is not an individual student problem, it is a systemic one that requires systemic tools to address.
The Deterrent Effect Is Immediate
Once students understand that AI use and copying can be identified, behaviour shifts. The value of detection is not only retrospective accountability, it is prospective cultural change.
Cumulative Insights Drive Better Teaching
The class-level analytics gave the lecturer a clear picture of shared student weaknesses (i.e., budget management, strategic justification, and framework application) that he could address proactively before the exam.
Scale Does Not Have to Mean Compromise
213 students. Two classes. One batch. Three days. The long-standing tradeoff between class size and feedback quality no longer has to be accepted.