image from work

Dreaming Ourselves Forward: The Sunnyside Wisdom Project

  • by

  • CrAIyon, Dream.AI
  • doi: https://doi.org/10.7273/4psh-1f24
  •  BEGIN 
Authors' Note: Using generative AI for content creation as well as pedagogical purposes can be fun, exciting, and creative, but it is not without its risks. The authors of this work and the people involved in the creation of the artwork approach generative AI and AI systems broadly with a critical lens, which includes serious conversations about equitable, diverse, and inclusive representation, labour, subjective experiences, as well as the responsible use of technologies. While the AI tools mentioned here–CrAIyon and Dream.AI–had great educative potential for students and are also free for use to the general public, they should also be explored for their impact on existing creative industries and practices, including through exploiting and impacting the work and livelihoods of human artists. The Dreaming Ourselves Forward project and its reflection below approach these vital unseen contexts of generative AI seriously. Our key objectives are awareness and human agency through pedagogy and knowledge exchanges with the intention to teach our youth that as they continue to encounter AI in their everyday lives, they also have agency in learning to responsibly engage with AI's affordances and limitations.

Dreaming Ourselves Forward: The Sunnyside Wisdom Project

Dreaming Ourselves Forward: The Sunnyside Wisdom Project is a multimodal, multilingual artwork composed of the individual perspectives of over 250 students in grades 7 and 8 at Sunnyside Middle School in Kitchener-Waterloo, Canada. Our project uses AI image generators to visually represent the “wicked problems” that are tackled by humanity but to which young people may have particular perspectives, interests, and stakes. Indeed, the adult decisions made today–about problems from regulating technological advancements to mitigating climate change–are the responsibilities that young people must contend with tomorrow. The student participants of Dreaming Ourselves Forward collaborated with adult instructors (Sunnyside teacher Graham Baechler, university professor Dr. Lai-Tze Fan, and soon-to-be Master's student Tatum Weicker) in using generative AI more responsibly and for social good, resulting in a final public-facing artwork.

We began our project by allowing the students to wrestle with several inquiries, including: What is wisdom? What truths are important to me right now in my life? How might I use what I know to be true right now to serve my future self? In addition to awakening students to their inborn capacity for wisdom and creativity, the project also seeks to function as a means of resisting and speaking back against self-limiting beliefs imposed by oppressive systems, including the education system itself. Since many of the questions that the students addressed in the final artwork are musings over and predictions about the future, generative AI lends an ironic, albeit aptly critical lens through its own predictive operations as it is designed to take training data–for us, the students’ prompts–and turn it into visions of and wisdoms for the future.

Designing AI Workshops for Middle Schoolers

In preparing for this project as both pedagogues and researchers, Lai-Tze and Tatum were careful to address concerns of privacy and safety, including to make sure they offered a workshop and technological literacy that could ensure that the students and their parents felt safe with the subject matter. Classroom-provided Chromebooks were used with monitored access to the internet. These real-world contexts and conditions were of high importance to Lai-Tze and Tatum as a way to extend their research practices to community-facing participation and work responsibly (Fan 2021a). For projects in research-creation (creative approaches to research), Lai-Tze is particularly interested in using artistic practices as a way of getting outside of academic environments and reaching out to extra-academic and non-academic audiences (Fan 2021b).

In preparation for co-creating the artwork, Lai-Tze and Tatum conducted an in-class workshop at Sunnyside Middle School in December 2023. They worked to determine which generative AI application would be best suited for the project, along with essential information for the students to understand the potential implications of generative AI. While the students ultimately used Dream.AI for the final artwork, Lai-Tze and Tatum focused on CrAIyon–a free AI art generator using the deep learning model DALL-E 2 to generate images from text user-input prompts. It had seemed the most accessible and easy-to-use application for the students to work with throughout the project. Lai-Tze and Tatum put together a slideshow presentation and workshop for Graham's classroom of eighth-grade students, giving an overview of what generative AI is and how to use CrAIyon as a tool to generate their own images that represent their individual ideas, outlooks, and opinions. There was a brief discussion about how the AI had been trained using collected data, as well as emphasis to the students that data can collected on popular platforms that include the social media tools that they are most familiar with. An important goal of the workshop was to establish a critical thinking lens: students were taught to not immediately accept AI results as truths but rather as representations to be critiqued and altered as necessary. By learning how to write and revise effective prompts, students used generative AI to create images that served as expressions of their wrestling with truth, a way of dreaming their future possibilities.

During the workshop, Lai-Tze and Tatum provided examples of the biases that can appear in generated content from initial prompts. For instance, they demonstrated an example of gender bias that results from asking for images of “nurses” from CrAIyon. The output predominantly showed white women. Afterward, they asked the students about the accuracy of the images, which came with immediate retorts that men could be nurses. The students were asked what they thought could be improved upon, resulting in suggestions for a more detailed prompt: “nurses exclude ‘women.'" Through this exercise, they saw that offering details in the prompts (input) could drastically change the represented results (output), which they suddenly learned to question.

Next, they asked for images of classrooms, which generated pictures of groups of children who were all light-skinned. A Black student noted that the students did not look like him, leading us to agree with him. They explained that one reason that the AI represented classrooms as containing white children is because it had likely been trained on examples of white children in classrooms. Afterwards, as a class, everyone tweaked the prompt until it displayed an image of a racially diverse classroom. Students participated in the workshop to understand how generative AI might be used as a creative tool and learning tool. The students were shown that adjusting their prompts helps to produce images that best suit their visions and ideas in response to practice questions.

The students were tasked to work on their own for ten minutes and their responses were inspiring. They had quickly adapted to the various versions that the image generator provided, discovering the necessary tweaks and details to add to their prompts that could more closely reach their vision. Each student seemed eager to use something new in the classroom, jumping into the task and happily sharing their results with their peers as well as with Graham, Lai-Tze, and Tatum. A small introductory workshop was enough for students to get excited about using generative AI as a tool to transform their words into images as a means of creatively visualizing and dreaming what was in their imaginations.

The Creation Process

After Lai-Tze and Tatum left the classroom, Graham decided to share the information gained from the workshop to each class in the school–grades seven and eight–through an adapted version of the workshop that he taught to all Sunnyside students. There were problems using CrAIyon for output and in terms of reliability so ultimately he used Dream.AI as the main AI art generator. From January to April 2024, Graham led students to explore their own individual critical lenses, inquiring into the experience and lessons learned for 12-, 13-, and 14-year-olds experimenting with generative AI art for the first time. He showed the students the idea of not accepting initial results from AI through also demonstrating the biases that can result from various prompts and language, allowing them to learn how to utilize it as a tool for their intended purpose.

To make the information more manageable for young students, Graham held a school-wide assembly sharing his own students' initial tests to allow the school to visualize what the process of AI-aided artist creation could look like and what outcomes may be produced with each individual prompt. Although including the entire school in this project added unaccounted layers of logistics, Graham believed in the project and wanted to make sure that each student’s voice and critical lens were seen and heard. He notes a positive reaction including from other teachers at Sunnyside. In order to work with their students he reflected on the need to respect their existing workloads, not wanting to over-stress or overburden them in their own programming.

Graham’s assembly was the real turning point for such an interactive project ensuring all faculty and students felt clear and comfortable in the process to produce the best possible results. Graham’s excitement and interest in their work was reflected onto students, giving an energy to the project that carried into its fruition, motivating young students to showcase their thoughts using a new medium. In the final work Dreaming Ourselves Forward around 250 Sunnyside students' AI-generated art and wisdoms are represented.

It was important for Graham to embrace an open-minded stance towards learning about AI since he wasn’t coming into it with pre-existing knowledge or expertise. He notes that forcing himself to be open to play and experimentation was also important to model for the students. As the project progressed, he and the students embraced the spirit of modifying and adjusting according to new information and common problems they faced along the way, whether it was learning new features of the AI tools or finding ways as teachers to get as close to the student learning as possible within the context of a whole-school project.

The Artwork

The students’ role in this project was to portray their individual wisdoms to the world, showing what the future looks like to 12-, 13-, and 14-year-olds in today’s age and dreaming themselves forward through their work. To “dream themselves forward,” Graham asked students to write or select a wisdom statement or meaningful quotation that resonated with them. The sources from which they selected their wisdom statement varied, including a documented list of student generated ideas from various learning experiences, religious texts, advice from family and elders, video games, and social media. Students were asked to reflect on why they were drawn to their chosen idea and why it mattered to them personally. Once they arrived at what is true for them at this point in their lives, they then considered how that truth could positively serve themselves five to ten years from now.

The final artwork is a beautiful range of imagery that holds wisdoms truly unique to each student. Students’ truths include themes of religiosity, valuing yourself whilst respecting those around you, patience, perseverance, and selflessness. Each piece is a wonderful representation and reflection of young students’ wisdoms. Using critical lenses and interpretations to apply new technology to a complex subject shows the creativity and creative adaptability of the students at Sunnyside Middle School.

Reflections

Critical Thinking through Ubiquitous Technologies and AI Tools

After the artwork was created, Lai-Tze and Graham sat down to reflect on how the experience had been for both as well as what they had both learned. What started as an e-mail from a stranger–Graham to Lai-Tze–became a conversation about how we as educators of students across age groups and levels of education can leverage our creative desires within the context of our jobs while also taking some risks that we may not otherwise take.

Our collaboration allowed students to get into the world of scholarly critical thinking at younger ages and to participate in relevant social dialogue about using AI tools. This participation includes more philosophical framing and wrestling with notions of truth (whose truth? For what periods of time? For how long?) as well as to contribute to those dialogues and feel out their places as young people in those dialogues. Indeed, the process of the project for Graham revealed many of the unique perspectives of the students he had been working with throughout the school year.

As students displayed increased interest in and engagement with learning how to use Dream.AI, Lai-Tze and Graham considered the suitability of the chosen tool for the age group. Technology now plays a central role in youth education–the Chromebooks that the students used were already provided by Sunnyside Middle School–and kids are used to interfacing with screens as a way to learn. In many cases, popular technologies can be a unifying tool of audiences across ages since digital tools such as smart phones and the internet–and increasingly AI tools–are designed to be ubiquitous in society and culture. For this reason, students across age groups (whether in middle school or university) as well as disciplinary backgrounds or foci (whether following exercises in computer class or pursuing a degree in Sociology) tend to approach media technologies with a sense of familiarity compared to other learning subjects and objects that they might approach for the first time.

However, this does not always mean that technologies are used in critical ways or for critical practice regardless of age group or education level. It remains important to re-contextualize popular tools for critical use, including to connect classroom conversations about emerging tools such as interactive and generative AI to students' digital lives outside of school. These technologies we wanted to stress to them are one and the same. How they use them can be more responsibly guided and minded.

Equity, Diversity, and Inclusion while using Generative AI

As equity-minded teachers, researchers, and artists, our personal practices and classroom practices center around equity and justice. Graham encouraged this mindset to his students as well as reflexively to himself through leading workshops, including by interrogating potential gaps in knowledge and asking: what does it mean for us to continue to build safe space and inclusive space? His questioning allowed us to imagine how having an entire school use the same tool in Dream.AI serves as a common denominator of experience, open play, and experimentation. In the same way, using the same tool allowed Graham to compare student experience and output as he identified intersecting commonalities as well as important, respected differences among the students.

In terms of inclusivity for student participants, we continued to consider the ways in which even popular and ubiquitous technologies may not be equitable in use or equally accessible for all potential users. For instance, historically in many countries worldwide, there have been lower numbers of girls and female-identified students who are interested in or encouraged to pursue STEM (science, technology, engineering, mathematics) topics compared to boys and male-identified students. In the workshops, we found that there wasn't a gender disparity in shyness or reticence for participation. Based on this reaction, we thought about what is unique to generative AI tools such as Dream.AI that allows for a low enough barrier to entry such that users, teachers, and students alike can encourage more of an inclusive and engaging experience. How we accomplished this was to continually ask students about their personal relationships to AI-generated content; if they were not satisfied with generated content, we followed by asking: what could be questioned in the tool and what could be potentially tweaked or fine-tuned in the prompt? Having a personal experience and negotiation with the tool allowed each student to understand its potential affordances and limitations, its benefits and risks.

Using Generative AI Tools to Augment other Classroom Topics

Following Graham's instruction on how to use Dream.AI, the conversations started to explore other capacities of the AI tool, including to amplify some of the learning processes in other subjects. For example, in the shared close reading of a poem, Graham asked his own students to highlight a line or an image that struck them or that resonated most deeply with them. The students made a selection and then were challenged about how to use or adapt the selection into an AI prompt. After trying their experimental prompts with Dream.AI, they further explored and discussed the resulting generated images as an expression of how the students connected to the poem or how the generated images added additional layers of meaning. In another example, the students used Dream.AI to visually represent a character in a story and had a discussion about which representations they accepted and which they rejected and why. The conversation about representing identity in a diverse way allowed students to question accepting default output by AI tools. Finally, Graham notes that using the generative AI tools with students felt like just scratching the surface and that he understands that to explore more possibilities with generative AI tool usage would require becoming more proficient with prompt engineering as a natural next step. To that effect, he asks what does good prompt writing and engineering look like for students aged 12-, 13-, and 14-years-olds? How can we sharpen and hone that as a writing skill?

Getting Kids to Slow Down: Mitigating Some Bad Habits in using Generative AI

Graham and Lai-Tze saw some of the negative habits and dangers of using generative AI too–namely that students and adults alike love the speed and instantaneous responses of our everyday technologies. When students wrote prompts for Dream.AI, used an interesting filter, and watched as an interesting image appeared in front of them, it sometimes worked against the critical thinking lens that we were trying to establish. Critical thinking requires us to slow down a little bit. Students needed intentional learning conversations to help them gain a clearer understanding of their purpose. Students benefited from questions such as “What were you trying to get at with this prompt?” and “How might changing the words in the prompt or trying a different filter change the results to better match what you’re looking for?” We found that slowing down the process at the right time increased the likelihood of students understanding how the choices they made were connected to the final image generation.

Opening Night

The physical art installation Dreaming Ourselves Forward was on display from June 3 – 14, 2024, at the Stanley Park Community Centre in Kitchener, Ontario, Canada. The installation ran alongside another called Teen and World Issues, led by the art teacher at Sunnyside Middle School, Chantry Gesinghaus. Graham attended the Opening Night with his daughter and parents. Lai-Tze and Tatum joined along with some of their friends and colleagues at the University of Waterloo to see the completed work in its totality for the first time.

Below are photographs of the installation from the Opening Night.

photograph of the installation from the Opening Night photograph of the installation from the Opening Night photograph of the installation from the Opening Night photograph of the installation from the Opening Night photograph of the installation from the Opening Night photograph of the installation from the Opening Night

Conclusion: Dreaming without End

In a concluding conversation between Graham and Lai-Tze, they reflected on the difficulty of keeping students–whether in middle school or university–open-minded to new ideas, tools, and ways of thinking. Lai-Tze reflected that it is often the case that as we accrue more education–through to high school and sometimes university and graduate school–it can be hard to unlearn certain information that becomes a safety blanket for adults. And sometimes it can be more difficult for adults than children to move away from what feels safe, known, and in control. It seems that the older we get, the less and less our society instills into us a freedom and feeling of safety in exploring the new, the unfamiliar, and the different.

Graham echoes that as we get older, there are certain behavioral patterns in our belief systems that get further and further entrenched. If one asks a child in kindergarten about the possible uses of a paperclip, they can come up with a laundry list. With older groups of students, there may be less and less experimentation and more subscription to knowing the use of a paperclip. This does this; that goes there. He describes this change in the mindset and exploration of his students as a cognitive rigidity–something Lai-Tze recognizes as well as the making hard and brittle of minds in how we encounter new information and ways of knowing, a calcification of creative exploration..

When we introduce AI tools and systems to our students–including and maybe especially students thinking through the humanities and social sciences who are self-professed non-experts and even a few Luddites–we are met with a unique opportunity to defamiliarize otherwise familiar devices like computers and smart phones. It is precisely for this reason that as a team, our interest in teaching critical approaches to AI across disciplines and age groups focuses on diverse perspectives pondering how to make human-AI experiences more equitable, accessible, transparent, and fair.

Here’s our hope for the 2024 cohort of students at Sunnyside Middle School, as for other youths who are interested in and confronting emerging technologies like generative AI: more than the hard facts, rules, and “known” ways of living in the world, may the sincerity of your childhood wisdoms be what will carry you forward.

Please live well.

LTF GB TW
June 2024


Works Cited

Fan, Lai-Tze. “Research-Creation for the Community: Pedagogy, Feminist Maker Cultures, and the Critical Work of Making Face Masks in the Time of COVID-19.” English Studies in Canada, vol. 44, no. 4, 2021 (listed as 2018 due to backlog), pp. 39-46.

“Critical Making, Critical Design” edited by Lai-Tze Fan. the digital review (creative contributions of double issue). 12 September 2021. Funded by the Canadian federal SSHRC Connection Grant. https://thedigitalreview.com/issue01/index.html.

SSHRC. Definitions of Terms. Social Sciences and Humanities Research Council of Canada, Government of Canada. https://www.sshrc-crsh.gc.ca/funding-financement/programs-programmes/definitions-eng.aspx#a22. Accessed 1 June 2024.

is in his 17th year of teaching Grade 8 at Sunnyside Senior Public School in Kitchener, Ontario, Canada. He seeks to empower students through literacy development with an emphasis on social justice. Working towards a more equitable society is central to both his personal and professional growth. He believes that good things happen when we take time to learn about how to best ensure all students feel welcome, safe, and valued.

is the Canada Research Chair in Technology and Social Change and Associate Professor at the University of Waterloo, Canada, as well as Professor II at the University of Bergen, Norway. She leads The U&AI Lab at U Waterloo, which intervenes in biased Big Tech design by creating alternative resources and methods for AI, with a focus on enhanced EDI outcomes. Fan is an Editor and the Director of Communications of the open-access journals electronic book review and the digital review. Fan is the Editor of recent special journal issues on research-creation, including 2021’s “Critical Making, Critical Design,” which won the ELO’s 2022 N. Katherine Hayles Prize for Criticism.

is an undergraduate student at the University of Waterloo, Canada, majoring in Sociology, minoring in Political Science and Human Rights. She is a forthcoming Master’s student in Sociology at Acadia University, Canada. She has previously done research on the disposal practices of solar panels in Canada and their harmful socioeconomic consequences. Her primary research interests are class-inequality, social movements, and neoliberal governance.