Deep Learning Assessment Must Not Be Punitive!
In conventional educational practice, assessment is often narrowed down to the act of assigning numbers, determining rankings, and then deciding who is considered successful or a failure. Consequently, assessment feels like a judge’s gavel delivering a verdict. Students who receive low scores are viewed as less capable, while teachers feel their job is done once the numbers are entered into the report card. This perspective is not aligned with Deep Learning.
In Deep Learning, assessment is not a tool for punishment but a compass for understanding the student’s learning position, improving the journey, and ensuring that learning objectives are truly achieved. Thus, assessment becomes a pedagogical conversation that helps teachers understand how students think and helps students see that learning is a process of growth, not a competition to defeat friends or merely an arena for finding fault.
Because it functions as a compass, assessment cannot be confined to the end of the learning process. It must be present from the beginning, continue throughout the process, and be used again at the end. These three timings are not just a technical sequence but a single cycle of information that helps teachers and students make conscious decisions.
Initial assessment is needed to map students’ readiness, prior knowledge, interests, experiences, and potential misconceptions. Teachers must not assume that all students start from the same point. This initial data serves as the basis for adjusting the level of difficulty, selecting pedagogical practices, forming study groups, or providing differentiated support. Initial assessment is not a test to label students as ‘clever’ or ‘weak’; it is a diagnosis to ensure the learning dosage is correct.
During the learning process, assessment functions to read development. Teachers can use open-ended questions, discussions, reflection journals, demonstrations, short quizzes, concept maps, interim project results, or observations of how students solve problems. This information must be immediately followed by feedback and strategy adjustments. Black and Wiliam (1998) demonstrated that formative assessment used to adjust teaching can lead to significant learning gains. Therefore, what matters is not the number of instruments, but whether the evidence of learning is genuinely used to help students move forward.
At the end of the learning process, assessment is necessary to see the level of objective achievement and the quality of student understanding. However, final assessment in Deep Learning must not merely test the ability to recall lesson content. Students must be given the opportunity to demonstrate whether they can understand concepts, apply them in different contexts, connect ideas, solve problems, produce work, and reflect on their learning process. A final grade can still be given, but the number must be accompanied by meaning: what competencies have developed, which parts are still weak, and what learning steps need to be continued.
Within this framework, there are three types of assessment: assessment of learning, assessment for learning, and assessment as learning. Assessment of learning is used to make judgments at the end of a unit, semester, or programme. Its function is important for reporting and accountability, but it must not be the only form of assessment. Assessment for learning is used to improve teaching; teachers gather evidence, interpret student difficulties, and adjust their instruction, while students receive clear feedback about goals, current achievement, and how to close the gap. Assessment as learning positions students as active agents who assess and manage their own learning. Through self-assessment, peer assessment, rubrics, and reflection, students learn to recognise the quality of their work and determine their next steps. Earl (2003) emphasised that assessment as learning strengthens students’ metacognitive abilities and self-regulation.
These three forms are not mutually exclusive choices. Assessment of learning provides a picture of results, assessment for learning improves the process, and assessment as learning builds learner independence. In Deep Learning, balancing all three is crucial because the goal of education is not merely to produce grades, but to cultivate conscious learners who understand their own progress and take responsibility for their learning process.
Errors in assessment can damage learning. If teachers only chase final grades, students will learn for the test, memorise superficially, fear making mistakes, and avoid challenges. If instruments are not aligned with objectives, what is measured is not the competency that should be developed. If criteria are not transparent, assessment feels unfair. If feedback is only numbers or demeaning comments, student motivation and self-confidence can collapse. Moreover, erroneous assessment results can lead teachers to permanently label students and lower their expectations of them.
Therefore, Deep Learning assessment must be authentic, continuous, fair, transparent, and educative. Student mistakes must be read as information, not a disgrace. Low scores must be a sign that help is needed, not a reason to punish. Good assessment does not stop at the question, ‘What is your score?’ It moves towards deeper questions: ‘What have you understood, what is still hindering you, and how can we improve?’ That is where assessment becomes part of learning, not a trial held after learning.