Overview
In our Master’s level Game Design Programme, we introduced ChatGPT to cultivate students’ critical thinking and deepen their understanding of the biases, limitations, and capabilities of AI tools in comparison to human creativity. I observed that students were occasionally producing similar answers in other areas of the curriculum, which raised the possibility of their reliance on AI tools for idea generation. Therefore, I tasked the students with using ChatGPT to generate game ideas and critically discuss these. This provided them with a comprehensive grasp of the role of AI in game design and encouraged them to explore the potential of AI in conjunction with human creativity within the game development domain. The set task enabled them to identify patterns in both human and AI responses and to devise strategies to harness the strengths of each source.
Our approach
The module Game Design Methods introduces a unique approach to encourage students to come up with game ideas that address the topics and methods introduced by the tutor and discussed in the classroom. Specifically, students are given weekly design briefs as homework and required to first develop their own response, before feeding the given prompt to ChatGPT.
Subsequently, all responses are discussed in class. The primary goal of this formative exercise is to raise awareness about the use of AI tools in game design and to demystify their application. By presenting both human and system-generated ideas, the task highlights the differences between human and AI-generated concepts.
Process:
- I provided the students with design tools and frameworks to think about the problem. In the first week, they received the design brief and a week to come up with their own idea. In this case, they were asked to think about a game related to the experience of being a Master’s student, including their ideas and expectations of what it means to be a Master’s student. The students were then required to flesh out their own game ideas, with some even producing simple paper-based design prototypes.
- After coming up with their ideas, they used the same wording from the brief as question prompt to ChatGPT. They were not allowed to follow up on or tweak the first AI response. As instructed, these responses were brought back to the classroom. They presented both ideas in class, and we discussed them.
- The ChatGPT game ideas predominantly focused on simulation games about the life of a Master student. There were many similarities in the ChatGPT generated responses, e.g. they suggested having as goals of the game: ‘passing degree’ and ‘achieving successful career’. Although one response identified: ‘passing degree’ and ‘achieving a good work-life balance’.
- The AI generated outputs did not provide guidance on how ideas and abstract concepts could be created and made tangible in games. For instance, one AI response included ‘study’ as a behaviour in the game but did not explain how this could be represented through game actions. In another week, when the brief was to design an “adaptive difficulty system” AI suggested games with complex procedural generation, whilst students thought of simple, clever tricks to achieve the effect. Additionally, all AI outputs were about digital games, even though the prompt had not specified this.
- One student’s idea was quite poetic and focused on life decisions, designing a game about using a real scale to physically balance activities such as ‘going to the pub’ or ‘reading a textbook’.
- In discussion with peers and me, students can identify and explore patterns in both sources and determine ways to harness the potential of each.
What benefits has AI had on our practice?
It is a little too early to state with certainty that students’ critical skills or their work has improved. I observed the following, however:
- Students could spot similarities and shortcomings of AI responses, fostering a deeper understanding of the strengths and limitations of AI in game design.
- Presenting both human and system-generated ideas meant that students were exposed to different perspectives and could explore the potential of AI alongside human creativity in game design.
- This transparent use of AI created an open environment for students to freely talk about their use of such tools.
- Students’ own game ideas often focused on player psychology, emotion, and reaction, seemingly encouraging complex game designs to research such behaviour and preferences, and thus enhancing the development of more engaging and immersive games.
- It has highlighted the importance of the conceptual tools and frameworks taught, as they help address game design problems in ways that AI didn’t seem prone to do at first.
What have we learned?
Despite the positives observed and the interactive nature of the tasks described above, students still seemed to have some difficulty connecting AI use in formative exercise to summative work (in the case of this module the summative assessment is a short ‘interactive’ essay about a game design topic which includes the creation of a complementary game). Students may require further/clearer signposting and explanation to combine the use of AI with design frameworks and to optimise output, e.g. to explore diverse game ideas and character portrayals. Students may realise that there is some bias in AI tools without realising the underlying reason for this. An inherent algorithm bias, for example, can lead to a tendency to generate similar stereotypical outputs. So, in short, AI might help students identify different answers, but such answers might all be built on the same stereotypes. The ultimate aim is to help students apply such tools more intelligently and critically.
“I think it is a great tool with a great potential. I am not sure if it is healthy or not, but it feels like a very important technology to me. Ethically, using AI feels like it’s wrong but this clearly won’t stop its quick development or people who are using it. So, I believe it is important to use this technology in my learning process to get ready for the future.” Oğuzhan Cetin. MA Game Design Student.
Oğuzhan Cetin (MA Game Design Student)
“It felt like we were using AI as the opposite of inspiration; instead of copying what it was generating we were basing our work on what it wasn’t generating. One of the things we noticed quite quickly was how vague and plain the ideas it was generating were – so it was great as a benchmark to make our ideas more interesting and unique. It was good in the classroom to compare our ideas against it and see what we knew vs. what the AI knew, as it gave us a chance to stay one step ahead of it.”
Holly Cooper (MA Game Design student)
Our future plans
I will continue to use and embed AI tools in my practice with students. One area to explore further with students is how to identify hidden patterns and training biases. Another is to train the software by feeding it personalised input to produce ideas that are more bespoke and personal to each designer’s style.
Currently, there are not many students on this module, which allows the discussion of each original idea and AI response. Further plans are to identify, design, and develop tasks that can be used for larger cohorts of students.
This case study has been captured by the Academic Development team and showcases current practice from MA Game Design (MAGD7102 Game Design Methods). We would like to thank Rafael Arrivabene (School of Art, Design and Architecture) for participating in this case study.