AI - A Threat to Students' Development?

Both drawings below are made by children. Which one would you want your child to give you for Christmas? For my part, I have no doubt - I would wish for the hand-drawn picture. Let's take a closer look at the two.

The drawing of the puppy is created by a child using generative artificial intelligence (GAI). The idea is fun and the drawing maintains (of course) a high technical level. The other drawing is also created by a child, but it is hand-drawn. The idea is not as clear, and technically it has far lower quality. Nevertheless, there is no doubt that I prefer the hand-drawn picture if I get to choose. Why is that? The three most important reasons are that the hand-drawn picture is

...personal.

• The child tells a story. We don't know if it's true or fictional, but most likely it springs from the child's own experiential world in some way or another: The people are drawn differently, the staircase, the window where you can see other houses outside. I don't quite understand the arrow on the right side of the picture. But since an arrow by definition is a symbol with a clear intention, I think the child has a thought behind this.
• In the AI-generated drawing, all elements have a generic design. The houses, snow-laden trees, Santa hat, the puppy with its head slightly tilted - everything fits perfectly into an idea of winter idyll. The only thing that might resemble a personal intentional expression is the very idea of a puppy sledding.

...relational.

• The child has drawn eight people, or, I choose to interpret it as four people in two different situations. We don't know who these are, but the context allows us to guess. It could, for example, be the child's own family. That's clearly an interpretation, but the child has nonetheless drawn people who are in a relationship.
• The AI-generated drawing contains no clear relationships.

...shaped by effort over time.

• Based on general knowledge of children's motor development, we know that the hand-drawing has required effort over time. This applies both in the moment - for example when lines have become a Christmas tree that then needs to be colored. At the same time, we know that the same child's drawing of a Christmas tree looked different a year earlier, and will continue to develop over time.
• To the extent that the AI-generated drawing changes, it is primarily a result of developments by the technology provider being used. The child may switch to another service or develop their linguistic instructions. However, that's something entirely different.

All these three - personal, relational, and effort over time - are linked to values that rank high in my personal value hierarchy. That's why the hand-drawing, in my eyes, is so incredibly much more interesting, despite the AI-generated one being clearer and with technical drawing quality of another world.

It's all about process

Why do we ask a 14-year-old student to write a text about democracy? I think everyone agrees that it is not because the world lacks texts about democracy. We obviously want to initiate a process in the student, with gathering, processing, and consolidation of knowledge, and reflections that contribute to the students' perception of the world. In addition, it's a meaningful way to practice expressing oneself in writing, forming text and/or using digital tools. Yes, and several other processes.

The same applies to draw, creating music, and other forms of expression. Of course, there may be contexts where it makes sense to use generative AI to create a drawing or a piece of music. But in most cases, neither the drawing nor the music is the primary purpose - it's the process of thinking and the development. This means we should think about the goal, and what might prevent us from reaching it, before we ask students to use generative AI in education.

The value of long-term, continuous processes

What are possible obstacles in the example with the two drawings above? As mentioned, it depends on the goal. We could talk about the student getting less motor training. We know this can have lasting ripple effects far beyond drawing. We could also talk about less training in aesthetic thought processes. While verbal language is symbolic, and thus relatively limited, aesthetics can express a broader range of emotions. That is one of the strengths of aesthetic thought processes, and something we should be careful not to remove.

In this case, however, I want to focus on something that may not be quite so obvious, but which I also have experienced as a music teacher:

The hand-drawing shows the student's development of motor drawing skills, creativity, and aesthetics in a long-term perspective. We're talking about processes that last continuously for at least 10-15 years, and often longer. The child in our example is perhaps somewhere in the middle of this journey.

With the AI-generated drawing, the student skips the process and achieves a much higher technical result, in an extremely short process. 

Can this weaken the sense of mastery in making hand-drawings? In my experience, yes, undoubtedly. Students who compose most of their music digitally can easily become dissatisfied with the technical level of their own manual process. This therefore decreases in value for the student, despite possibly having a much higher artistic value in terms of composition, sequence, theme, expression, and originality. 

This is because it takes training to endure training. We get this type of training from when we are infants. No one learns to walk by being told how, but by observing, trying, and failing over a long period.

What may happen if education - with generative AI - removes this type of long-term processes and simultaneously increases expectations for technical level?

One possible consequence is that students lose the experience of developing through effort over time. When technology can deliver impressive results in seconds, patience with one's own learning curve may be weakened. This can affect students' general attitude toward learning - not just in arts, but in all subjects where development requires time and practice.

One thing that can prevent us from reaching the school's goals is thus that long-term development and training decreases in value for students. It's not given that this will happen, but it can happen. Awareness and good teaching help a lot, and that's the point: Awareness of possible obstacles can help us use generative AI in appropriate ways. Therefore, we must constantly ask questions, not just about what is possible, but also about what might prevent us from reaching the goals.

Suggested actions

How will you handle this type of complex challenges in the classroom? And how can we ensure that awareness reaches not only teachers with high digital competence, but broadly into all classrooms? My preliminary suggestions for actions are:

• Acknowledge the value of long-term processes
• Reflect on the difference between creating a good prompt versus creating a drawing
• Take extra steps to identify possible consequences of using generative artificial intelligence
• Let students practice seeing what technology can offer and what they themselves must contribute as humans

In this text, I have focused on challenges of using artificial intelligence to generate products, exemplified by drawing. I have not addressed how AI can be used as support in a creative process. We'll have to return to that.