Teaching machines

December 9th, 2021

by Matteo Loglio


One of the things that we enjoy doing is teaching and lecturing about our work, and we've been doing that for quite a while in schools and events all around the world.

In this article I want to talk about some of the workshops we've been running over the last few months. A recap of the last year in teaching. Why is it so important to talk about these activities? Well, since the nature of our practice demands constant exploration of uncharted creative territories, spreading this knowledge is equally valuable as the work itself.

An oio workshop at HEAD Genève

In the last two years while teaching in various schools and institutions all around the world, we’ve been experimenting with a new workshop format called Design with Other Intelligences. This involves collaborating with a class of art and design students (and professionals) for a solid week, guiding them into experiencing firsthand some of the oio magic✨ process, creating artifacts and products in collaboration with an algorithm. How does the design process change once we add another intelligence to the human team?

AI Architecture by Ilhana Besic and Raghid Jomaa at HEAD Genève

The core nugget of the workshop is for participants to experience a conversation with the algorithm, the Centaur principle - half human, half machine.

The core nugget of the workshop is for participants to experience a conversation with the algorithm

The process is not linear - it’s not as simple as feeding the program with heaps of data, then publishing the results. We have to go one level deeper, taking the output of the AI and use it to re-design a new artifact, so that the final outcome of this exchange between humans and machines is not just the output of the algorithm itself.

Here’s a couple case studies. In October I was teaching a short course at the Master of Media Design at HEAD Genève while in November I was lecturing a similar course at FABRICA, Treviso .

The process

In both workshops participants had to pick a subject for their project: posters, pictures, typesets, consumer products, text, music, basically any human artifact that could be fed to a machine learning algorithm in the form of pixels or text. After choosing their subject they would select a Generative Algorithm based on the input - StyleGAN for images, GPT for text - then train it on the dataset they created. After the machine generates new outputs, participants would take these to design radical new artifacts, then finally produce all the assets to communicate the idea.

Over the last few years I've been using mostly RunwayML to teach this kind of workshops, and it's great. Runway is basically a machine learning playground, lowering the entry barrier to students but also artists and designers, as it requires no code or previous knowledge. It's not free but they provide free credits and education licenses which are perfect for these activities, and the team is very friendly.


Here below some projects from the students.

↑ This is AI-KEA Balloons by Emma Parente, Melina Reymondin and Clément Vogelsperger at the Master of Media Design at HEAD Genève. Here students trained a machine learning algorithm using a mix of IKEA furniture and balloons, getting a mix of shiny inflatable furniture for your living room. After getting those images, they modelled the chairs in a 3D software like Blender and created a fictional IKEA catalog.

↑ The project above is called Fictive Islands by Pauline Baldinetti and Andreia Rodrigues at the Master of Media Design at HEAD Genève, who worked on fantastic landscapes dreamed up from Google Maps imagery of remote islands.

These islands don't exist

In the project below called Posts&Pots, Audrey Délisse and Théotime Serre scraped thousands of influencers pictures from Instagram, trained an AI to generate new images, then they sculpted a vase AND 3d printed it. You can see some steps from the process below.

This is the outcome of the algorithm...

...that students used to model a vase and 3D print it!

Here below some projects developed by the creatives at FABRICA, the communication research centre of Benetton in Italy.

FABRICA in Treviso

The following images come from an abandoned church, where Melissa Maarek, Anastasia Miseyko, Alessia Lapio and Pietro Bucciarelli created AI Madonna, an installation with algorithmically-generated Madonna statues were displayed with generated music as well. Together with the installation they also created AI-generated prayer cards ("Santini") that visitors could take home.

AI Madonna in the abandoned church

Algorithmic santini

While in this last project called Portrait of a street Bobbi Fay, Noah Ringrose, Evelin Mazzaro and Dorian Etienne trained a generative algorithm on their hometowns, generating new glimpses of invisible cities and using them to model new artchitectural elements. Here below is a model of a bridge, inspired by the algorithm.

When algorithms dream of cities

Yes I've been using the word "algorithm" eleven times in four paragraphs

What's next?

The way we run these activities varies from an exploration of just a few days, all the way to a week-long experience. We can’t run too many of these at the moment as we are a small team, but the good news is that we are expanding (check out our careers page) and this is certainly an important area that we’d love to keep growing.

We are still running workshops in schools, events and private companies too. If you are interested in exploring the future of creativity together, let us know by dropping an email at hello@oio.studio, or join our Discord to find out about all the upcoming events.

Stay smooth 🍎🐛