Turn a PDF lesson into a quiz: a teacher's editing and fact-checking workflow

To turn a PDF lesson into a quiz, select the pages that support one learning objective, generate a small draft and check each key against the original page. Readable extracted text can still lose a label, condition or unit. Review those details before students receive the questions.

This guide uses a six-step workflow and a worked science example. It covers readable source material, question selection, fact checking, feedback and student data. You can use the same sequence with a textbook excerpt, a worksheet you own or your own lecture notes. Keep only questions whose sources, answer keys and feedback you can verify.

Start with one lesson objective, not the whole file

Imagine a six-page PDF on plant nutrition. It includes diagrams, a definition of photosynthesis, an extension on limiting factors and an unrelated homework instruction. Your next lesson only needs students to explain why light matters to photosynthesis. Do not ask for a quiz on “everything in this PDF.” Write the objective first: students can identify light as a necessary condition in a simple investigation and explain what a result would show.

Now select the two pages that support that objective. A narrow input limits the chance that a generated quiz spends three questions on an appealing but irrelevant sidebar. It also gives you a manageable source to verify. If the PDF is a scan, inspect the extracted text: a missing minus sign, misplaced column or unread diagram can change the answer. If a diagram carries the key idea, add your own description or write that question manually.

The UK Department for Education’s guidance on generative AI in education says generated material may be inaccurate, out of context or low quality, and that the professional remains responsible for the final resource. That is a useful rule for a lesson PDF: treat the output as editable teaching material, not as an authoritative answer key.

Generate a draft with a specific brief

Ask for a small set of questions, not a maximum count. A useful brief names the class level, objective, permitted source pages, formats and what to do when information is absent. For the plant example, try a request like this:

Create five low-stakes questions for students who have just read pages 2–3 of this lesson. Check whether they can explain the role of light in a simple photosynthesis investigation. Include two multiple-choice items, two short-answer items, one explanation question and a reason or marking criteria for each key. If the source does not establish a fact, do not invent it.

The last sentence matters, but it cannot guarantee correctness. Open the original PDF alongside the draft. For each question, find the passage, diagram or teacher explanation that supports the key. If you cannot find it, remove the question or supply a checked source yourself. Our broader guide on generating a quiz with AI explains the types of input and output; here the focus is the review between generation and publication.

Skolina’s AI quiz workflow can take a PDF or lesson content as a starting point. Use it to reduce first-draft typing, then edit questions and feedback before sharing. The step-by-step guide to an online classroom quiz shows how a completed quiz is delivered to students; the source PDF itself is preparation material, not something students must upload.

Check every answer against the source

Review the draft in a fixed order so that polished wording does not distract you from a wrong answer. A question can sound completely reasonable and still rely on a claim the PDF never made. Use this table as a quick decision guide:

In our example, “Plants get food directly from soil” could be an informative wrong option if you have seen that misconception. “Plants make food from moonlight” is merely odd. A stronger item asks students to predict what happens when two otherwise similar plants receive different amounts of light, and to name what else must be kept the same. The answer now exposes reasoning rather than recall alone.

The guide to writing good multiple-choice questions offers more detail on plausible distractors and clear stems. You should still read all options yourself. Automated generation does not know which misconceptions your students actually hold, and it may introduce phrasing that gives away the answer.

A worked PDF extraction audit: preserve labels, numbers and units

For a small pilot, use a teacher-authored example that you can inspect directly. Suppose page 2 states: “Plant A receives six hours of light daily; plant B receives two. Each receives 50 mL of water daily.” Page 3 contains a fictional height table: both start at 10 cm; A finishes at 16 cm and B at 12 cm. The document states the same variety, growing period and other controlled conditions. These numbers illustrate the editing workflow; they are not experimental evidence about a species.

The third row is a useful trap for the editor. With equal starting heights, 16 − 12 = 4 happens to equal the growth difference. If a later version starts A at 10 cm and B at 11 cm, their growth becomes 6 cm and 1 cm, so the difference is 5 cm, not 4. Check the method in the feedback, not only whether the current key is numerically right.

Save a source record with the checked draft

Record the file version, selected page numbers, the relevant caption or cells and the final key in your own preparation notes. If the PDF changes, recheck the affected questions rather than assuming the old source location still supports them. A colleague should be able to reconstruct 6 − 2 = 4 cm from the document without relying on the generated explanation.

Skolina generation uses one Token per generated question, with file-extraction costs in addition; inspect the current plans and limits before budgeting a file-based batch. Content submitted to AI generation is processed by OpenAI outside the EU, while account data is hosted in the EU. Use permitted lesson material without identifiable student data, then preview the final quiz separately from the PDF: a readable original does not guarantee readable question images.

Protect the document and the people in it

Before uploading a PDF, inspect its contents. A teacher-authored lesson handout is different from a document containing student names, marks, accommodations or private correspondence. Remove identifying information and use only the material necessary to create the quiz. Check your school’s approved-tool policy and the service’s data terms before uploading any protected document.

The UNESCO guidance on generative AI in education calls for attention to data privacy and human validation in educational use. The Department for Education’s school data guidance likewise recommends checking whether a tool is approved, understanding how it uses personal data and fact checking results. These are process checks; this article is not a substitute for your school’s privacy assessment.

Check copyright and permissions as well. Owning a copy of a textbook PDF does not automatically give you permission to upload or republish it through every service. When uncertain, work from your own summary or an excerpt you are allowed to use. You usually need the learning objective and a few relevant paragraphs, not the entire book chapter.

Edit feedback and test the student view

For each closed question, add an explanation that helps a student repair a misunderstanding. “Correct: B” is weaker than “B, because only the light condition changed while the other conditions were held constant.” For an open response, write a brief rubric or example of the reasoning you expect, but read actual responses before deciding what they show. Do not assume that a generated sample answer covers every valid explanation.

Before you publish, take the quiz yourself on the student view. Confirm that a link or QR code opens, images are readable on a small screen, the answer key is correct and the feedback appears at the right time. If a question depends on a diagram that was lost during extraction, include the diagram in the quiz or remove the item. See the Skolina pricing page for current plan options if you are planning a recurring workflow across classes.

Keep the final set short. Five reviewed questions can check one objective in a few minutes; twenty unreviewed questions can consume lesson time and create false confidence. Save the best item for reuse and revise it after you see student answers. The result is not merely a quiz made from a PDF. It is a teacher-checked assessment built for a learning decision.

A six-step checklist for the next lesson

Use this sequence when the next PDF lands in your folder:

1. Define the learning decision: what should tomorrow’s responses tell you?
2. Select the relevant pages: remove unrelated text and personal information.
3. Generate a small draft: specify the audience, formats and source boundaries.
4. Verify each answer: find the supporting sentence or diagram in the PDF.
5. Rewrite weak options and feedback: make common errors visible and explain the correct reasoning.
6. Test before sharing: open the student link and inspect the quiz on a phone.

If you want to try the workflow with your own lesson, create a Skolina teacher account, start with two relevant pages and publish only the questions you can verify against your source.