But a grammar quiz that adapts to your mistakes is much more useful.
Instead of asking AI for twenty random grammar questions, ask it to behave like a diagnostic quiz creator.
Start with:
“Create an interactive English grammar quiz for me. Ask one question at a time and wait for my answer. Cover tenses, articles, prepositions, question formation and subject-verb agreement. Do not show the answers before I respond.”
Then add an important instruction:
“If I answer incorrectly, don’t immediately move on. Give me a similar question to check whether the mistake was accidental or whether I actually don’t understand the concept.”
Now the quiz is doing something different.
It is testing your understanding.
You can make it more personalised by specifying your level and goals:
“I am an intermediate English speaker and I want to improve my spoken English. Focus on grammar mistakes that commonly affect everyday speaking. Avoid obscure grammar rules.”
This prevents the quiz from becoming an academic grammar examination.
You can also ask AI to identify patterns.
“Track my mistakes throughout the quiz. At the end, group them by grammar area and tell me which two areas need the most practice.”
For example, you might discover that you got most questions right but repeatedly struggled with questions and articles.
That gives you a direction for practice.
You can then ask:
“Based on my quiz results, create a ten-minute practice session for my weakest grammar area. Make me produce my own sentences instead of only choosing answers.”
This is where assessment becomes useful.
The quiz itself is not the destination.
It leads to practice.
You can also create different quiz formats.
For example:
“Create a grammar quiz using sentence correction only.”
Or:
Or:
“Create a grammar quiz where I have to complete short conversations.”
“Test me by giving me spoken-style sentences. Ask me whether they sound correct and why.”
That last format is particularly useful for spoken English because real communication does not arrive as multiple-choice questions.
You can even make the quiz progressively harder:
“Start at an intermediate level. If I answer three questions correctly in a row, increase the difficulty. If I make two mistakes in the same area, slow down and give me more practice.”
AI is especially useful here because it can change the experience based on your performance.
And there is one final instruction I recommend:
“At the end, do not simply give me a score. Give me a diagnosis and three specific things I should practise next.”
A 16/20 score is information.
“Your biggest weakness is forming questions in the past tense, and you need to practise this through speaking rather than more multiple-choice questions” is a plan.
That is the difference between testing and feedback.