Open-book assessment is no longer a small exception reserved for a few university courses. Schools and universities are testing versions of resource-supported exams, while generative AI has made take-home and open-resource assessment harder to separate from everyday study. The important learning question is not whether students can look things up. It is whether they have practiced enough retrieval, organization, and error checking to use resources under pressure.
One visible example is the International Baccalaureate’s Diploma Programme open-book exam pilot. The IB says the pilot is exploring exams in which students use permitted resources such as books, summaries, notes, formula materials, or websites, with the goal of letting students focus more on problem-solving, critical thinking, and application. The pilot involved about 270 schools worldwide and included subjects such as literature, economics, and psychology. Read the IB overview of its DP Open Book Exam pilots.
The details matter. The IB’s economics pilot allowed a limited booklet of real-world examples, but not definitions, explanations, analysis, or evaluation. Its psychology pilot allowed a teacher-generated booklet of study summaries, but not a ready-made explanation of the concepts. In other words, the resource was meant to reduce the burden of remembering every detail, not replace the student’s responsibility to understand, select, apply, and justify.
Higher education is facing a similar issue through AI-supported assessment. A 2026 arXiv paper, “Reimagining Assessment in the Age of Generative AI: Lessons from Open-Book Exams with ChatGPT”, described engineering students who were allowed to use ChatGPT during take-home open-book exams and had to submit their interaction transcripts. The author reported that stronger evidence of learning appeared when students tested outputs, corrected incomplete responses, and justified decisions, not simply when they obtained a final answer.
That is also why some instructors are moving in the other direction. The Associated Press reported that more college professors are using oral exams and other live assessment formats to check what students can explain without simply submitting polished AI-supported work. See the AP report on colleges turning to oral exams in response to AI.
What The Research Adds
Recent learning-science research gives learners a practical warning: access to materials does not automatically change preparation in a useful way. A 2025 study, “Test preparation: the roles of expecting access to materials and test anxiety”, examined how expecting access to a book or notes related to preparation and test anxiety. The authors reported that anticipated access to materials did not reliably change anxiety, and the relationship between anxiety, study time, and note production was more complicated than the simple idea that open-note testing makes preparation easy.
At the same time, retrieval practice remains strongly relevant for open-book tests. A 2025 open-access study in Learning and Instruction, “Effects of retrieval practice on retention and application of complex educational concepts”, found that retrieval practice helped learners retain and apply complex research-methods concepts, especially when students completed enough retrieval practice and were tested after a delay. That matters because open-book exams often ask students to apply ideas to unfamiliar problems, not merely recognize a definition on a page.
A 2026 perspective in npj Science of Learning, “Trends in testing effect research: from lab to classroom, but not yet for all learners”, also emphasized that active retrieval tends to improve learning more than passive review, while noting that real classrooms need attention to individual differences and support. For resource-supported exams, that suggests a balanced approach: learners should practice recall and application before opening the book, but teachers should also teach students how to use permitted resources strategically.
Why It Matters
The risk of open-book exams is not that resources exist. The risk is that learners mistake resources for readiness. A student who has only highlighted notes may know where information is located but still struggle to decide which idea applies, explain why it applies, and avoid a misleading answer choice or attractive AI response.
The opportunity is just as important. A well-designed open-book or take-home exam can reward the kind of thinking that matters outside school: finding relevant evidence, comparing cases, applying a principle, checking a claim, and explaining a judgment. But those skills depend on having knowledge available in memory. If every step requires searching, the learner runs out of time and working memory.
This is where retrieval practice, concept maps, and error review fit together. Retrieval practice builds usable access to the core ideas. Concept maps organize relationships so the learner can navigate a topic quickly. Error review turns wrong answers into a search plan: what concept was missing, what cue was misread, and what kind of problem should be practiced next.
The Practical Learning Conclusion
Before a resource-based test, study as if the first attempt is closed book and the second attempt is open book. The first attempt reveals what you can actually retrieve. The second attempt teaches you how to use resources to check, refine, and repair your thinking.
- Start with a blank-page retrieval check. Write the main ideas, formulas, cases, or theories from memory before opening notes.
- Build a concept map after recall, not before it. Put the central idea in the middle, add causes, examples, exceptions, and applications, then check the map against the source.
- Practice finding the right resource quickly. Use tabs, headings, or an index, but rehearse where information lives so exam time is not lost to searching.
- Use practice questions in two passes. First answer without help. Then open resources and mark what changed: missing fact, weak connection, wrong procedure, or poor interpretation.
- Review errors by category. Separate memory errors from concept errors, reading errors, calculation errors, and evidence-selection errors.
- Prepare for transfer. After studying one example, ask how the same principle would look in a different case, graph, passage, dataset, or scenario.
- Treat AI like an allowed resource only when policy permits it. If it is allowed, use it to challenge your reasoning and generate practice, not to replace your first attempt.
The main lesson from recent open-book exam news is clear: resources can support better assessment, but they do not remove the need for memory. Learners still need enough retrieval strength to recognize the problem, enough structure to choose the right resource, and enough error-review habit to notice when a polished answer is wrong.