What happened
Recent education news is sending a consistent message about assessment: schools are under pressure to see more of the learner’s actual thinking, not just the finished answer. That matters for academic integrity in the age of AI, but it also matters for learning. A good formative assessment does not simply catch mistakes. It helps a learner notice the mistake soon enough to correct it.
On August 10, 2026, the New South Wales government said it had asked the NSW Education Standards Authority to review the impact of AI on student learning and consider a moratorium on unsupervised take-home assessment tasks for the Higher School Certificate. The announcement says the review will examine AI’s effect on cognitive development and how assessment methods, including major projects and creative works, can better safeguard authentic student work. Read the official announcement here: NSW Government to review AI impact on student learning.
A few days earlier, Denmark moved in a similar direction for upper secondary schools. According to The Guardian, Danish pupils aged 16 to 19 will have to orally defend written essays as part of a response to AI-assisted cheating, and schools are being encouraged to shift more written work into supervised conditions. Read the report here: Danish pupils will have to orally defend essays in attempt to combat AI cheating.
At the classroom level, the same issue appears in a less dramatic form every day. Teachers need quick evidence of what students understood, what they guessed, and what they can explain without help. A March 2026 NWEA article collects practical formative assessment strategies such as exit tickets, quick writes, misconception checks, and short evidence-gathering routines. Read it here: 27 easy formative assessment strategies for gathering evidence of student learning.
The common thread is not only “stop cheating.” It is “make thinking visible.” When assessment becomes only a polished product, a score, or a submitted file, learners can miss the most useful part of the learning cycle: seeing the gap between what they thought they knew and what the task actually required.
Why it matters
Formative assessment is valuable because it happens while learning can still change. A final exam may certify performance, but a short quiz, oral explanation, draft check, worked step, or exit ticket can show the learner what to do next. The timing is the point.
The recent AI assessment debate sharpens this point. If a student can outsource a take-home answer, the teacher may lose evidence of the student’s current understanding. But the learner also loses something: the productive friction of trying, getting stuck, revising, and choosing a better strategy. That friction is where self-correction develops.
Oral defenses, supervised tasks, and short in-class checks are not automatically better just because they are harder to outsource. They are better when they force useful evidence into the open. Can the learner explain why a method works? Can they identify which step failed? Can they compare a first answer with feedback and revise it? Can they transfer the idea to a slightly different problem?
This is where formative assessment connects with retrieval practice. Retrieval practice asks learners to bring information to mind instead of only rereading or recognizing it. RetrievalPractice.org summarizes the learning principle here: What is retrieval practice?. A low-stakes quiz or explanation prompt gives learners a memory check and a misunderstanding check at the same time.
The practical risk is that schools respond to AI by making assessment more controlled but not more useful. A locked-down quiz can still be too late, too vague, or too focused on points. A better response is to build frequent feedback cycles in which students try something, receive specific information, and immediately act on it.
The learning conclusion
The practical conclusion is simple: formative assessment should end with a next move. If a quiz, draft comment, AI-tutor hint, oral defense, or teacher conference does not help the learner decide what to correct next, it is only partial feedback.
A useful self-correction cycle has four parts:
- Attempt first. The learner answers, solves, writes, explains, predicts, or recalls before looking at the model answer or asking for help.
- Compare with evidence. The learner checks the answer against a rubric, worked example, teacher comment, peer question, source text, or answer key.
- Name the error. The learner identifies the type of gap: missing fact, wrong procedure, weak explanation, misread question, careless step, or poor strategy choice.
- Choose the next practice step. The learner does one targeted action, such as reworking one problem, rewriting one paragraph, explaining one concept aloud, or practicing a narrower prerequisite skill.
For teachers, this means designing checks that reveal the source of an error, not just whether the answer was right. Ask students to show one step, defend one choice, mark their confidence, explain a misconception, or revise a wrong answer after feedback. Short, frequent checks usually beat large, delayed corrections because students still remember how they were thinking when the mistake happened.
For learners, the habit is equally practical. After every quiz or feedback moment, write one sentence: “My next practice target is…” Then make it specific. Not “study algebra” but “practice isolating the variable when there are fractions.” Not “improve writing” but “revise topic sentences so each one makes a clear claim.” Not “read again” but “close the book and explain the three causes from memory.”
For parents and tutors, the best question after a result is not only “What grade did you get?” It is “What did this show you to practice next?” That question turns assessment from judgment into direction.
The news from NSW and Denmark shows why education systems are rethinking assessment in the AI era. The learning takeaway is broader: students need assessment moments that make their thinking visible, give timely feedback, and require a concrete correction. The goal is not just a more secure assessment. It is a learner who can notice an error, revise the strategy, and choose the next useful step without waiting for someone else to rescue the process.