Slow AI

Abstract The algorithm has become the enemy: Our daily lives are attention-harvested by recommenders, feeds slopped by generators, grocery prices gouged by segmentors, and speech censored by classifiers. Hype has pushed the pendulum too far, and now it’s come crashing back. Fear, anger, burnout, and sheer economic precarity reign. The pendulum slams hard, back and forth, again and again. Automate any job, no humans needed! *swing* Boycott all automation, the technology is evil! *swing* Optimize every millisecond of your day! *swing* Let me be unproductive! *swing* GenAI is the democratization of art! *swing* AI is a parasitic theft-machine! *swing* There’s too much energy in the system. Time to slow down to the deliberate, messy pace of the human bodymind. This is Slow AI. Slow AI, the culmination of the 2025 Leonardo CripTech AI Lab, is a virtual exhibition that serves as both sanctuary and provocation. Come rest with us as we Crip the algorithm. Slow AI stands with the Slow Food and Slow Fashion movements while also rooting into the slow reality of Crip methodologies. Our method used community care to make technical choices: open-source, local, private models, along with Crip rage, creative access, and the friction of misunderstanding. The virtual exhibition 1 stages a journey through a digital landscape. The sun has set. We are wandering along a quiet creek. There are boulders where we can stop to rest. We move slowly. We think and speak slowly. There is no rush. Visitors explore a series of digital installations that showcase the work of nine visionary artists who have reimagined the possibilities of machine learning. The works span multiple technological eras, utilizing techniques ranging from 1950s cellular automata to contemporary large language models (LLMs), and involve various fields and technologies, from mapping spaces with Crip bot bodies and prosthetics for hearing-impaired musicians to speculative medical imaging and bio-data sonification. Including both solo-authored and collaborative works, these projects focus on the expertise of disabled lived experience, accessibility as a creative framework, and emerging technologies as a site for Crip intervention. In Mat Dalgleish’s limp.freq, a simulation in which bots with leg-length differences map the accessibility of virtual worlds by painting the ground with their feet, snail trails reveal the world’s hidden hurdles and unreachable corners. Lee Kuo-Ying’s (李國瑛) Complementation (互補), a visual prosthetic for musicians with sound sensitivities and hearing loss, shifts signals from flinching ears to hungry eyes. Zsofi Valyi-Nagy’s Feeling is Believing, a pair of video and radio ads for a speculative medical imaging machine inspired by aura photos, exposes the need for self-knowledge unserved by the cursory glances of the medical industry. Abram Stern’s Enhance-Redact (Cancer Surveillance Study #1), a loop of enlarging, reducing, and interpreting medical imagery smaller than a pixel, is an endoscope that surveils the theater of battle, be it tumor or warzone. In Movement in Squares, Lucia Grossberger Morales stages a recursive collaboration between the human hand and algorithmic intelligence— a choreography in which brush dances from pixel to text and back. Through Cripping LLMs, interabled collective Christian Bayerlein, Yesica Duarte, and Puneet Jain invite difficult and poetic conversations using a custom LLM designed to dig into beliefs about disabled bodies. Slow AI is Crip rejection of the “frictionless” future promised by Big Tech. It proves that the algorithm does not need to be an immutable force of extraction but can be a malleable material that can be slowed, broken, and remade in our image. These digital landscapes reject metrics of growth or engagement in favor of being in community with complexity, rage, and profound care.