<aside> 📍 This page shares updates on the Small AI stream within the bigger Slow AI research project by AIxDESIGN.


<aside> 🏄‍♀️ The research on Small AI is being led by Nadia Nadesan.

Nadia is a fellow at Open Future and a collaborator at Platoniq with experience as a researcher, illustrator, facilitator, urbanist, and technologist. Her work focuses on civic technology, data governance, and public space. As a founder of Design Justice Mediterranea, she is actively involved in creating collective spaces of resistance, knowledge sharing, and community. She has worked as a researcher under Akwugo Emejulu on the project ‘To Exist is to Resist’. Her latest project with Algorights at the Jornadas DAR developed citizen modes of participation in AI governance facilitating a citizen assembly around the AI Act in Europe. Her research currently focuses on the materiality of artificial intelligence and frictions where technology encounters local sociopolitical and cultural contexts. She enjoys projects and networks that encourage re-learning, reorienting, and reclaiming narratives of who and why we are, especially how we tech and eat.




01_ Preface

<aside> 🐜 TLDR: Small AI questions the dominance of large-scale models by addressing environmental impact, discriminatory language, and cultural preservation. It’s an invitation to use notions of fractals, friction, and fragmented-ness in nature and math to inspire ideas around Small AI and send ripples that shift practices around Big AI. It advocates for equitable, sustainable alternatives, prompting exploration of community-focused governance, and interconnectedness as pathways to more just futures.

In this research we ask: What is possible when our objective is cultural creation and preservation instead of scale or profit?



What is Small AI?

Small AI confronts the pressing issues raised by scholars like Timnit Gebru's question of whether language models can indeed be "too big”. Building off of the work of earlier scholars it considers the risks such as the environmental toll of massive AI models, the reproduction of discriminatory language and biases, the particular focus on manipulating language versus preserving or designing around the integrity or cultural specificity of a language, and the potential for AI to propagate misinformation. With an understanding of the harms and risks of training large AI models, Small AI helps us conceptually and practically understand how we might cultivate and create new relationships with technology, nature, and culture.

small ai.png

Small AI – a Challenge to Big Terminator Narratives

Small AI questions the narratives around larger industry standards and practices. Consider that 2023 was the year of AI hype. Generative AI exploded on the scene with Chat GPT’s new capabilities of producing even more plausible text. Then image generators like Stable Diffusion, MidJourney, and DALL-E also entered the public discourse with incredible images generated from just text prompts. But the constant growth and expansion of generative AI’s capabilities comes at the cost of cutting corners as an industry standard. Each AI corporation relentlessly racing to out do each other have resorted to plagarising the works of artists and creatives to grow their training data and dismissing the cost of creating training models on the planet. In parallel to this ‘growth’ Sam Altman like Beyoncé was on tour lobbying politicians and public officials about AI regulations both in the United States and Europe. His tour was almost as dramatic as the legendary pop diva with his dire warning that AI could kill us all. Small AI challenges the overtly Oppenheimer notion that AI must inevitably mushroom into a colossus expressing humanity’s drive to innovation and its simultaneous self destruction.

Innovation is not about size or power. Confronted with the very real perils of the current climate crisis and the extractive nature of AI industry practices, another world has to be made possible. Small AI is about rethinking the narratives and ecologies around technology where people and the planet are not a footnote in the story of technological progress but at the center. A new narrative prompts a pivotal question: what possibilities lie in embracing smaller-scale AI initiatives? By shifting the focus away from relentless scalability and profit-driven models, Small AI presents an opportunity to re-imagine technology's relationship with communities and ecosystems. This approach aspires towards meaningful engagement, cooperative ecologies and economies and nurturing life-affirming systems that prioritize community well-being.

The Case of Te Hiku Media

Smaller initiatives like Teh Hiku Media provide an example of what values and principles motivate design that considers the well being of people and the place they live. Te Hiku Media is a media organization ‘collectively belonging to the Far North iwi of Ngāti Kuri, Te Aupouri, Ngai Takoto, Te Rārawa and Ngāti Kahu’ with the mission to inter-generationally instil, nurture, and proliferate the Māori Language.