Generating Access: Alt(er)text
25 July 2026, EWF x Next WaveIn this workshop session, Access Lab & Library (Jon Tjhia and Fayen d’Evie) will encourage participants to engage with networked culture through the framing of access. Why is alt text a contested space? How can we sustain critical positions in the mediated platforms we use to communicate? And how can practises like image description extend and expand the craft of writing?
Drawing on discourses around agency, authorship and knowledge creation, and through conversation, solo and collaborative practise, you’ll undertake a round of exercises that query authorship, speak to position and perspective, and recognise the generative significance of emergent knowledge.
Access this page
Or open:
https://accesslab.world/generating-access-participant-page
Session outline:
- 2.15pm—2.30pm: A mini lecture
- 2.30pm—3pm: Exercise
- 3pm—3.15pm: Discussion and reflection
Prompts (for you, not an LLM):
- How can alt text represent what is visually present in the image?
- How about what is in the image, but is not visually present?
- Can it describe negotiated or oppositional positions against the image?
- And how does auto-generated alt text reinforce ocularcentrism?
Exercise
You may work solo, or in groups of two or three.
We would like you or your group to enact one of these approaches for each of the images:
- Replace the generated alt text with your own
You may describe what is seen, or add unseen context. A further sentence may treat the alt text as a channel of intimate publishing; a subtextual or counter narrative, breaking the fourth wall, speaking directly to your audience - Supplement the generated alt text with a public/visible image description
Imagine that it’s publicly readable by all visitors to the site. If the alt text is there, how might you use the image description space as a place to express a particular writing position — and/or to provide a broader context? Do you read any potential relationships or shared themes across/between the four images? - Devise as many strategies to thwart, spoof and evade machine learning training models as possible
What are your counter-strategies? What are you good at that computer vision is not? Can you trick y/our way out of this bind?
Draft and share your responses here: https://pad.riseup.net/p/generating-access-keep
‘Three performers wearing reflective gold protective suits stand in a circle at an industrial composting facility, their connected sleeves extended between them in a sculptural formation. Mounds of compost and green waste surround the group beneath a blue sky with scattered clouds.’
‘Side-profile photograph of Pauline Hanson with eyes closed, wearing red lipstick and a white drop earring against a black background.’
‘Loose black-and-grey ink portrait of Gina smiling, with shoulder-length dark hair and a pearl necklace against a white background.’
‘Two tall paintings show crowds of stylised figures lifting pale nude bodies beneath a blue sky, creating a densely layered scene of intertwined people and gestures.’
TTYL!