[ONLINE] CAPTCHURED: Training Sets for Refusal

​CAPTCHA is a type of security test used in computing to determine whether the user click clack clocking is human or a bot. CAPTCHAs act as gatekeepers, deciding who gets access based on specific criteria, usually visual recognition skills. This mirrors societal rules, where institutions define who is allowed to participate based on compliance with often invisible norms.

​Instead of clicking through boxes to prove we’re human, what if we could invent our own tests of belonging? CAPTCHURED is partly walking, partly making participatory workshop. Participants will gather images from the surrounding environment to form personal datasets, then transform these into CAPTCHAs that resist recognition. The resulting works are invited to contribute to CAPTCHURED, a participatory artwork and growing archive of subversive puzzles that resist verification. Come walk, come make, come trouble the algorithmic gaze. 🐸

​Before we meet: undertake a 30-minute asynchronous walk 🚶⚠️

​Sometime before the workshop, we invite you to undertake a 30-minute walk and create a small personal dataset along the way. The walk can happen anywhere you choose preferably on the planet Earth: around your neighbourhood, on your way to work, through a park, somewhere completely unfamiliar, or wherever you happen to be.

​As you walk, take photographs of things that catch your attention. There is no particular category to look for and no need for the photographs to be beautiful, unusual, or even particularly interesting. What you notice is the dataset.

​Keep the photographs with you, you’ll use them to make your own CAPTCHAs during the workshop. You won’t be asked to upload or share the original dataset.

​🗺️ If taking a 30-minute walk isn’t possible for you, interpret the exercise in whatever way works from where you are: sit somewhere, look out of a window, move through a building, or find another way of spending 30 minutes deliberately noticing your surroundings, be it on the internet. : )

​Together we’re gonna ponder:

​1. What does access mean when machines decide who gets in?
2. Who is seen, mistranslated, or filtered out by machines?
3. Who gets to say what makes someone human?

​What are we gonna do?

• ​reflect on our own experiences of being verified / surveilled / excluded / filtered by digital systems

• ​[OPTIONAL BUT HIGHLY APPRECIATED] 30-minute walk outside while creating our own dataset by collecting weird signals, tldr: taking pictures in the wild

• ​play with CAPTCHAs as a form of language and power structure through the web tool specifically developed for this workshop

• ​archive and contribute visual squares into the ongoing CAPTCHURED artwork

​What are we gonna unlearn?

​In this workshop, we are unlearning the neutrality of CAPTCHA, which reduces complex identities and ethical nuances to yes/no answers, the inevitability of algorithmic borders, and the notion that machines should define what it means to be human.

​For whom?

​The workshop is open to curious individuals of all genders, backgrounds, and fields. No prior knowledge of AI or hardcore tech is required. In fact, it is enough if they have ever used the internet for more than 5 minutes in their lives.

​Deets

​📅 3 September, 17:00 BST - 18:00 CEST - 20:00 GST - 12:00 EDT - 09:00 PDT
💻 Online (Jitsi, Miro)
🫰 There is no ticket price for this session. Instead, if you're able, we invite you to acknowledge the labour of love behind it by contributing whatever amount feels fair to you. Your support helps us continue creating spaces for collective experimentation and resisting the machines. ❤️‍🩹

​The capacity is limited so please if you cannot make it, let us know or de-register! <3

​Institute of Machine Unlearning

​The Institute of Machine Unlearning is a para-academic performative infrastructure for collective unlearning, feral pedagogy & serious play. Practicing the weird third thing, between AI hype and doomerism, in a messy middle space of participatory inquiry through participatory workshops, strange tools and DIY publishing.