Background
In a session for MSc Data Science and AI, I gave a lecture on unconventional techniques of using generative AI models, such as circuit bending and glitching neural networks, to create unexpected artistic expressions. As a part of the learning outcomes, students will do a hands-on programming activity to gain practical knowledge of the techniques. The main challenge of this activity was approaching these advanced practical techniques while ensuring students with less strong practical skills could also engage with the activity.
Evaluation
In the activity, I used an interactive programming notebook as a task sheet (link). In the notebook, I provided step-by-step guidance for students to follow. A typical interactive notebook in programming pedagogy requires students to execute the “code blocks”, and eventually adapt part of the code in the notebook to complete a few tasks as elaborations.
I have experience in running hands-on programming activities over the past two years. A critical aspect of ensuring students with different levels of practical skills all engage with the activity is to (i) have straightforward instructions in the notebook for those who may require more help getting on with the tasks, while (ii) giving a high degree of freedom to explore for those who are looking for more challenging tasks.
Given these considerations, I planned and experimented with the following strategies to address the challenge.
Moving Forward
Converging points during the activity
In my previous experience, after students had signed off to work on the notebook, the classroom diverged to individual or small-group structures, and they worked at their own pace. However, for this session, I experimented with a diverge-converge-diverge strategy (Palmgren-Neuvonen et al., 2021): students diverge to work through the first half of the interactive notebook. Then, before entering the advanced sections, we converge and meet again in a lecture style to reiterate key techniques they have encountered and explain the following advanced sections. Then we diverge again to keep working individually.
In the actual session, this strategy was implemented smoothly. As a result, students were actively engaging with the advanced section of the notebook. For future applications of this strategy, the class size might be a consideration. The class I ran was relatively small, with around 10 students. However, in another MSc unit I’m assisting on, the class size can go up to 80 students. Getting the attention back in such a large classroom might be harder. Clearer instructions on where and when to converge might be needed.
Co-created board for results-sharing
The idea of results-sharing is a commonly used pedagogic practice in creative coding, and it has been used by colleagues at CCI (Fiala et al., 2016). Borrowing this idea, I set up a collaborative Miro board for students to communicate their results. The board allows students to upload their creations and the code they have written.
This strategy ensured students who worked smoothly throughout the notebook kept engaging in the classroom to discuss their creations. The collection of results was very interesting, as shown below (permission and consent to share these results were granted by students).
Group works/ group discussions
If I were to do the lesson again, toward the end of the session, I would encourage students to work in groups, discuss their results, and encourage students who have finished the tasks to help out students who are still working on them. The aim is to foster peer learning – less-experienced coders may seek help from their peers rather than face the challenges alone.
References
- Fiala, J., Yee-King, M., Grierson, M., 2016. Collaborative Coding Interfaces on the Web, in: International Conference of Live Interfaces. REFRAME Books, pp. 49–57.
- Palmgren-Neuvonen, L., Littleton, K., Hirvonen, N., 2021. Dialogic spaces in divergent and convergent collaborative learning tasks. Information and Learning Sciences 122, 409–431. https://doi.org/10.1108/ILS-02-2020-0043