How to Create Tests for Any Subject Using AI

Before we enter the world of artificial intelligence in education, we need to understand why current solutions are ineffective.

• To create tests manually: It is extremely time-consuming, prone to repetition (we often ask similar questions from year to year) and makes it difficult to differentiate tasks for students with different needs.
• Ready-made tests from the Internet: They rarely coincide 100% with the material carried out in lessons, they are substantively questionable, and their graphic quality often leaves much to be desired. In addition, they are easily accessible to students, which raises the risk of downloading.

We need a solution that combines the speed of a ready-made material with the precision and personalization of a hand-made material. That’s what AI offers for teachers.

AI as your personal test designer: How does it work in practice?

Stop thinking of AI as something complicated. Imagine that you are talking to an extremely competent colleague-teacher who has the entire core curriculum and hundreds of ideas for tasks in his head. You give the orders, and he does the work. The process of creating the perfect quiz or quiz is completed in a few simple, revolutionary steps. In our guide “Artificial Intelligence for Schools and Teachers” , we discuss specific tools and advanced techniques, and here we present a universal path.

Step 1: Foundation – instant generation of basic questions

Let’s start with the simplest and most time-consuming part – creating a pool of questions.

Example – History test for grade 5, section: “Poland of the Piasts”

Your prompt for AI:

“Create a set of 15 questions for the history test for grade 5 in the section ‘Poland in the times of the first Piasts’. The test should contain 3 types of questions:

• 5 single-choice closed questions (ABCD).
• 5 true/false questions.
• 5 open-ended questions, requiring a short answer (1-2 sentences).” 

Result (in less than a minute):
You get a ready-made set of questions, logically organized and ready to use. For example:

• (ABCD) Who was the first historical ruler of Polish? a) Bolesław the Brave, b) Mieszko I…
• (P/F) The congress in Gniezno took place in the year 1000.
• (Open) Name two reasons for Mieszko I’s baptism.

Most importantly, you can request to generate a response key, which will drastically reduce the time it takes to check in. You have saved at least 45 minutes.

Step 2: Differentiation on demand – Tests for each student

This is a real revolution in the approach to assessment. In the traditional model, everyone writes the same test. Thanks to AI, you can create two or three versions of the test adapted to different levels in a few moments.

Your prompt for AI:

“Take the previously generated test about the Piasts and create two additional versions based on it:

1. Version A (facilitated): For students with learning difficulties. In open-ended questions, add hints in parentheses. In closed-ended questions, limit your choice to 3 options (ABC).
2. Version B (extended): For gifted students. Add one question with an asterisk, which requires analysis of the source text (e.g. a fragment of the chronicle of Gallus Anonymous) and one task consisting in the chronological order of events.” 

Result: You have three ready-made, consistent versions of the test. Each student works to the best of their abilities. This is not only fairer, but also more motivating. The implementation of artificial intelligence in schools in this way really improves the quality of education.

Step 3: No more cheating – Unique kits for each group

The problem of dishonesty on tests keeps many teachers awake at night. Printing two groups (A and B) is standard, but AI allows you to go one step further.

Your prompt for AI:

“Based on the basic test about the Piasts, create 4 unique sets (groups A, B, C, D). Change the order of questions and the order of answers in closed questions. In open-ended tasks, slightly change the content of the question, preserving its meaning (e.g. instead of ‘Name two reasons…’ write ‘Give two reasons…’).”

Result: You have 4 different versions of the test, which makes it practically impossible to simply write back from a colleague. For math problems, AI can generate the same “types” of tasks, but with different numerical data for each student, which is an absolute game-changer.

Step 4: Testing Deeper Understanding – From Facts to Analysis

A good test should not only test memory. It should test the ability to think, connect facts and draw conclusions. AI is a great tool for creating tasks that explore higher cognitive levels.

Your prompt for AI:

“Create 3 tasks for the Grade 6 Nature Test from the ‘Ecology’ section that go beyond a simple definition.

1. Analytical task: Present a brief description of a fictitious forest ecosystem and ask the student to identify producers, consumers, and destructors within it.
2. Problem problem: ‘The number of algae in the lake has increased rapidly. Suggest two possible causes of this phenomenon and one way to reduce it.’
3. Graphical interpretation task: Generate a simple diagram of a food web and ask a question about the effects of the extinction of one of the species.” 

Result: You are given tasks that make students think. You check not only WHETHER they know, but HOW they can use their knowledge in practice. This approach to assessment, supported by intelligent tools, is the future of education, which we promote in our AI guide for education.

Step 5: Creativity and Visuals – A test that is not a wall of text

Who said a test has to be boring? By leveraging AI image generators and creativity in formulating tasks, you can create tests that are more engaging.

Example – English quiz, present tense

Your prompt for AI:

“Create 5 quiz tasks from the Present Simple and Present Continuous times. The tasks are to be creative. One of the tasks should be based on image analysis. Also generate a description of this image for the AI generator: ‘Comic-book style illustration of a park on a sunny day. In the foreground, the boy is eating ice cream, in the background the couple is running, and the old man is reading a newspaper on a bench.'”

Result: In addition to the standard loopholes, you are given the task: “Look at the picture and describe what each person is doing using the Present Continuous tense”, and you can also print a unique illustration.

Your Role as a Teacher: Verifier, Editor, Expert

Does this mean that AI does all the work? Absolutely not. Artificial intelligence is powerful, but only an assistant. Your role as an expert educator is crucial. It’s you:

• You verify the substantive correctness of: AI can be wrong, especially in niche or highly complex topics. You always need to check the generated content.
• You adjust the language and tone: AI writes correctly, but you know what language your students speak and what will reach them.
• You make the final decisions: It is up to you to decide which tasks best meet the didactic objectives and which will be included in the final version of the test.

AI takes 90% of the reproducible work off you, allowing you to focus on that 10% of expert work where your experience is irreplaceable.

Smart Grading Means More Time to Teach

Creating tests and quizzes no longer has to be synonymous with a sleepless night and frustration. Artificial intelligence puts teachers in the hands of a tool that allows them to generate high-quality, varied and engaging tests in a fraction of the time we have spent on it so far. It is the technology that frees up our most valuable resource – time that we can spend on working directly with the student, building relationships, improving professionally and resting on our own.

Ready to get started? Build your first smart test today!

This article is an introduction to the revolution in assessment. If you want to learn specific, tested tools, master the art of writing effective commands (prompts) and see dozens of ready-made examples that you can copy and apply to your lessons, our original guide “Artificial Intelligence for Schools and Teachers“ is made for you.

How to Create “AI-Resistant” Questions in the Chatbot Era?

Let’s talk about the fear that paralyzes many teachers from using open-ended questions. What good is it if we create a brilliant analytical task when a student can generate an answer to it on their smartphone in 10 seconds? This is a real challenge. However, instead of giving up tasks that develop thinking, we need to learn to design it in a new, intelligent way. An authoritative article has to face this problem.

The key is not to fight technology, but to outsmart it, by asking questions that require more than just information – they require personal experience, a synthesis of knowledge from lessons and authentic reflection.

Shifting the Accent from “CO” to “HOW” and “WHY”

• AI-prone question: “Describe the causes of the outbreak of World War I.” (The AI will generate the perfect, encyclopedic answer.)
• “AI-resistant” question: “Choose one of the causes of the outbreak of World War I that we discussed in the lesson, and explain why you consider it the most important. Use at least one argument from our class discussion to support your position.”

The latter question requires the student to make a personal assessment, refer to a specific event (discussion in class) and formulate their own opinion. AI has not been to your lesson and does not know what arguments were made there.

Leveraging Unique Classroom Context

The best defense against generic AI responses is to embed tasks in the unique context of your class.

• AI-prone question: “Analyze the figure of Antigone.”
• “AI-resistant” question: “During our debate about Antigone, arguments were made both for her attitude and for Creon’s reasons. Referring to two specific arguments put forward by your classmates, write which position is closer to you and why.”

Such a task tests not only the knowledge of reading, but also the ability to actively listen and synthesize. It is impossible for AI to do.

Evaluating the process, not just the final product

Instead of giving one big task, break it down into stages and evaluate your progress.

“Prepare an essay plan on… [tu temat]. Today, in class, I only evaluate the logical structure of this plan.”
Then, on the test, students write an essay based on a plan you have previously prepared and evaluated. This reduces pressure and prevents you from generating last-minute homework.

How does AI help with constructive feedback?

Creating and checking the test itself is only half the battle. Real learning begins when a student receives feedback – not just a grade, but a hint of what they did right, where they made a mistake, and how they can fix it. Writing detailed comments manually for 30 students is not feasible. And this is where artificial intelligence in education opens up completely new possibilities.

Generating a Comment Bank

After checking a few papers, you notice repeated mistakes, e.g. in a math problem, students notoriously confuse a circumference with a surface area.

Your prompt for AI:

“Many of my students make the mistake of confusing a square with the circumference of a rectangle. Create 3 different, constructive and empathetic comments that I can write for them in a notebook. One should contain a simple analogy (e.g., a perimeter is a fence and a field is a lawn), another should point to a specific step they have forgotten, and the third should include encouragement and a suggestion for additional exercise.”

Result: You get a ready-made bank of comments that you can quickly copy and paste, providing students with much more than just a crossed-out score.

Creating personalized “Repair Packs”

Imagine that a student had a problem with a specific type of task.
Your prompt for AI:

The student failed the Past Simple time test. Create a short, personalized ‘repair pack’ for them, including:
(1) A link to a short YouTube video that explains this time in a simple way.
(2) A set of 5 additional exercises in the form of filling gaps.
(3) One creative task, e.g., ‘Write 3 sentences about what you did yesterday using the Past Simple tense.'”

Providing a student with such a package is a powerful signal: “I see where you have a difficulty, and I give you concrete tools to overcome it.” This is the essence of formative assessment, which becomes possible with the support of AI.

AI in Everyday, Formative Assessment

Contemporary pedagogy moves away from the perception of assessment as a big, stressful event at the end of the section. The key is to collect information about student progress on an ongoing basis. AI for teachers is an ideal tool for creating quick, informal forms of knowledge testing.

Instant creation of “Exit Tickets”

Want to quickly see what students remembered from today’s lesson?
Your prompt for AI:

“I just finished a lesson on photosynthesis in grade 6. Create a short ‘exit ticket’ with 3 questions:

1. One closed question, checking the key fact (e.g. what is the product of photosynthesis?).
2. One open-ended question: ‘Explain in your own words why photosynthesis is important for humans.’
3. One reflective question: ‘What part of today’s lesson was the hardest for you to understand?'” 

The answers collected from students are invaluable information for you on what to pay attention to in the next class.

Designing Quick Diagnostic Quizzes (Pre-Tests)

Are you starting a new department and want to check what students already know about it so that you don’t break open doors?
Your prompt for AI:

“I am starting a section on Romanticism in Polish literature with grade 8. Create a quick diagnostic quiz (5 questions) in the form of an online survey (e.g. in Google Forms) that will test their basic knowledge of the era (knowledge of time frames, key characters, associations). The questions should be light and engaging.”

Such a quiz allows you to perfectly match the first lessons to the actual level of knowledge of the class, which is the foundation of effective teaching. It is this strategic approach to the entire assessment process that we promote in our AI for Education guide.