Playful and Creative AI Workshop in Freiburg

Last week I partnered with a local high school, the main branch of the public library and NEXUS experiments group at the University of Freiburg to prototype a new AI-based tinkering workshop. The project is part of a larger collaboration with the University of Connecticut and regional partners in both America and Germany who are working together to develop “emancipatory AI” experiences that help learners engage with new technologies, build a sense of agency and identify personally meaningful next steps.

The workshop took place during a special “project week” in the school schedule where students from 6th-9th grade changed out their usual day for this experimental workshop. The group that registered for the session happened to be entirely made up of boys which was a little disappointing since the workshop goals include creating new pathways for all students to engage with these tools and technologies. For the next time we’ll need to redouble our efforts to promote the event and develop a registration system that ensures more equitable gender representation.

On Monday morning, after a quick introduction to the workshop and facilitators, we got started with the idea of tinkering with AI and incorporating physical materials into our digital explorations by animating 3D collages using Meta demo lab tools. This idea came from a workshop with Sonia Tiwari and I really like how it feels playful but also opens up the idea that the predictive AI can have trouble understanding the “inputs” from the users and the prompts or information given to the model may have to be refined.

For the second half of the day, we explored the scratch lab face sensing tool as a way to create new interactive AI-based projects. As well, we introduced the idea of generating AI sprites and backgrounds for the sketches. The group took a bit to get familiar with this scratch extension but after a little playing around some really interesting projects began to take shape. A common exploration was a sort of fortune teller program that cycled through something like designer baseball caps, IQ scores or marvel superhero costumes before randomly assigning one to the face that the extension detected on screen. At the end of the day we did a little gallery walk around the lab to see what everyone made. For this project we used the pre-made scratch program blocks but I could also imagine another forking path in the workshop here where we could delve into training predictive AI much like the examples shown in Andy Forest and crew’s book about AI robotics.

For the second day of the workshop we started at the Freiburg Stadtbibliotek for an introduction to an experimental tool that I’ve been exploring along with the Scratch team at MIT Media Lab. In this unreleased version of scratch, a chatbot is directly integrated into the blocks so that you can generate characters speech and responses using a large language model similar to chatGPT or copilot.

We thought that this could be a good jumping off point for projects that use custom characters to share ideas and invite questions about topics that students care about. And the end of the morning session, the librarians also showed some resources like 3D printers that can be used in conjunction with AI tools to scan and print models in new ways.

Back at the school computer lab, we started integrating physical materials into the projects with Makey Makey, cardboard and foil switches and other craft materials to create new ways of interacting with the programs. These elements added another problem space to the project (in addition to the already large sandboxes of scratch programming and prompting the AI). I’m still not sure how much freedom or constraints make sense for this project. I would like to offer as many pathways as possible for exploring different aspects of the theme, but there’s always the risk of having too many options making things overwhelming for participants.

In a similar vein, I have questions about what works best for the format and locations of the workshop. For us, each day was hosted by a different partner (the school, the library and the university robotics lab) and we sometimes even moved between settings in a single day.

This was a really cool experiment and collaboration that exposed students to a wide variety of local institutions that are interested in exploring AI in educational settings. But the process of moving between different environments made it a bit hard to know what to expect in terms of technology connections and capabilities. As well, we lost some momentum with the more focused work of the projects. Maybe we could balance this more in future projects with a bit less movement between physical locations.

Overall, we had about 2-3 hours spread over two days to work on the AI projects and get a first taste of possibilities for the participants. All in all, I was impressed by the explorations and ideas that they worked with in the time available and I think there are many great starting points for continued investigation. Some examples included a sketch that shared information about favorite football teams with the user being able to toggle between FC Freiburg and Bayern Munich.

One interesting direction was a trivia generating frog that changed appearance and response if the user insulted it using pre-selected keywords. I think this concept could lead to some really deep explorations if we had more time to refine the parameters.

There were also several games with AI generated content like trash talk between cat and mouse or rewards of “interesting facts” every time the player reached a new level. One of the groups made a hilarious school simulator using an uploaded picture of their teacher interacting with themselves in a classroom setting.

The three days of this workshop were fun, challenging and generated lots of new questions and ideas to try in future iterations of aorkshops investigating AI in a tinkering context. Taking a zoomed out view on this topic, I’m still conflicted about the role of AI as a theme for a tinkering workshop or a tool to use while working on creative and playful projects. On one hand I have the sense that it’s important to engage with new technologies and build a sense of agency as an active participant in the physical (and digital) world. But I’m still worried about how AI might short-circuit the exploratory mode at the beginning of a project, lead to less individualized outcomes and flatten the scope of potential creations. And that’s not even considering the environmental impact and potential catastrophic predictions by experts in the AI field. As I work through these thoughts, I return to the belief that the best way to develop ideas around a new topic is to keep trying things in the real world with learners so that we can reflect about the implications of the projects and ask more questions together.

As part of the HRRC ‘emancipatory AI’ project, it was important to take specific platforms for projects like Scratch where the learning goals center around developing agency, playfulness, collaboration and human qualities as a jumping off point. I’d like to think more about issues of ethics and group norms related to generative AI topics and think about how more abstract concepts like democracy or human rights can be integrated into the workshop design without being heavy handed or instructionist. It’s been interesting to compare different scales of interactions from a drop-in celebration at the “AI Odyssey” event hosted by UConn to our targeted three day workshop to an ongoing afterschool program where learners have free choice to show up and work on personally meaningful projects at the YouMedia Center at Hartford Public Library. I feel that getting many different types of interaction and scale of participation helps us create a wide variety of examples of different possibilities.

I’m looking forward to doing more research and trying more prototype workshop experiments (next up is the inter.aktion festival at Experimenta in June) while grappling with the opportunities and challenges of creating AI tinkering experiences.