Beyond Prohibition: Levels of Artificial Intelligence Use in Schools
In schools, the first reaction to technical novelties is often prohibition. You cannot use the thing, period. This was the case with calculators, the internet, mobile phones, and, of course, artificial intelligence. This is particularly interesting because these tools are quickly gaining a role in the world outside school, while the school should prepare children for this life. This contributes to the school curriculum and usable knowledge moving further apart from each other, turning the school into a kind of *reserve* for skills required by previous technical eras. Artificial intelligence is revolutionizing the world faster than ever before, making different kinds of intellectual abilities important today than were yesterday, and it is likely that the next few years will also redraw the map of skills needed for success. As events are moving so quickly, it might be worth thinking ahead about how to regulate the use of artificial intelligence in school work, to which we provide a well-usable framework. Most institutions, even if they regulate the use of artificial intelligence in some way, think in terms of regulations that should be generally applicable to every situation, whether it’s using AI as a source and the mode of source indication, or the draconian rigor of enforced prohibition. In our opinion, pedagogical processes are much more diverse than this.
With the advent of the internet, one of the biggest problems seemed to be the issue of copyright. How to publish something on the internet so that anyone can use it, even process it, but the original author remains the owner of the rights. A simple and effective solution has been Creative Commons, which identified what could be done with intellectual property, and then identified different levels of usage based on these. It turned out, therefore, that it is much more effective to approach the issue in a more colorful way than just black and white. This is also the case with the use of artificial intelligence, where it is possible to use it on several levels, and it is advisable to decide which one to use for each task. We have identified five different cases:
DO NOT USE IT! In such a task, artificial intelligence cannot be used in any form, neither for brainstorming, searching for information, summarizing sources, nor for coding. In such tasks, with the submission of the work, the student also declares that they did not use any artificial intelligence application.
USE IT FOR CHECKING! In this case, the student should first solve the task completely without the use of artificial intelligence, and then ask the artificial intelligence if the solution meets the task. If there is a precise task description and evaluation table, the artificial intelligence can check the task based on that. If it points out errors, the correction should also be carried out without the use of artificial intelligence. In this case, the solution should be accompanied by what the artificial intelligence said about the task.
USE IT FOR BRAINSTORMING! In this case, before starting the work, the student can use artificial intelligence to generate ideas. After that, they should work out the task alone. The submitted work must also include what the artificial intelligence initially responded.
USE IT FOR DEVELOPMENT! In this case, the student prepares an initial version, shows it to the artificial intelligence, and asks it to improve and correct it. Whether they accept the suggestions or not is, of course, up to them. In this case, they can ask the artificial intelligence’s opinion on the work multiple times. The final submitted task must include the responses received from the artificial intelligence.
USE IT AS AN ASSISTANT! The student can use artificial intelligence to perform subtasks such as searching for sources, designing, summarizing sources, sketching, collecting data, and checking. The submitted work must also indicate which subtasks the artificial intelligence helped with.
USE IT AS AN AUTHOR! In this case, the student primarily works as an editor. They assign the task to the artificial intelligence and then suggest corrections. The student does not write but tries to steer the artificial intelligence towards a better solution. Along with the final material, they must submit the intermediate versions and how they directed the artificial intelligence.
If we discuss these rules, we can use artificial intelligence more safely while students also learn about its use. As an example, we describe a few situations in which the use of various levels may be justified.
When my students make diagrams, Excel only comes out after a while. The first bar chart must be made on graph paper in the classical way, and the first pie chart also in the traditional way with a compass and protractor. Excel’s excellent chart wizard only comes out after that, precisely so that it is not just a wizard, but so that the students understand what and why they are doing with it. For the same reason, the DO NOT USE IT! level might be justified.
It’s a cliché that feedback is the most effective pedagogical tool — the more detailed feedback students receive, the more effective the learning process will be. This is where artificial intelligence can also help, making the feedback even more thorough and detailed. It will not replace teacher feedback, but it can help produce better work, which is naturally important for detailed and well-usable evaluation guides to be prepared for the tasks. In such cases, the USE IT FOR CHECKING! rule can be useful.
Most often, artificial intelligence comes up as being excellently usable to overcome the *fear of the blank page*. In our experience, it is not really good for this because, although we get ideas, they are particularly mediocre. Students do better if they spend a few minutes gathering their thoughts and then ask artificial intelligence to expand on those. This can be used in tasks where students already have some basic knowledge, in which case it might be worthwhile to implement the USE IT FOR BRAINSTORMING! rule.
The main difference between brainstorming and development is at what stage of the work we want artificial intelligence to play a role, thereby what type of knowledge we want students to acquire. In the use it for brainstorming case, artificial intelligence provides the starting point, so we want the student’s work to be more related to the processing of knowledge, which is more similar to what is important in the workforce. However, if the goal is for students to gather basic knowledge, basic information, and then let the machine handle the processing, then the USE IT FOR DEVELOPMENT! approach should be chosen, as it more closely resembles what happens in school exam situations.
Most likely, in the near future, we will mostly use artificial intelligence in a way that the USE IT AS AN ASSISTANT! version allows. For students to do this well, it is important that they have basic knowledge of the topic and that they can identify those subtasks in which they can use the help of artificial intelligence. In such cases, just like with a calculator in math problems, the technology allows students to solve more and higher-level tasks.
At first glance, it might seem that the student’s role is not very large in the USE IT AS AN AUTHOR! case, but this is what requires the most complex activity. They must understand the generated solution, be able to identify what is wrong or needs correction, and be able to formulate this. This approach should be used when the student has practice in these thinking skills and has adequate knowledge.
These different versions can, of course, also be combined, for example, a task where artificial intelligence can be used both as an assistant and for evaluation, or where it can be used for development and brainstorming. If we specify with each task how this technology can be used, students will be more cooperative, and we can better focus on the areas we want to develop.
