AI, Design and Critical Imagination: Nadia Piet at Domus Academy

Piet opened by reflecting on the idea of “AI imaginaries”: the collective images, metaphors, and stories through which society understands artificial intelligence.

According to Piet, many of these imaginaries are shaped by big tech companies, whose influence extends beyond infrastructure, capital, and technical talent. As she noted during the talk, “Silicon Valley’s most powerful monopoly may be how we perceive technology.”

This matters because the way AI is represented affects how it is adopted, regulated, designed, and trusted. When AI is described as inevitable, magical, objective, or all-powerful, people may feel they have little agency in shaping it.

One of the key patterns Piet addressed was the dominance of big tech narratives. She challenged the idea that AI development is a neutral or unavoidable process, instead encouraging the audience to ask who benefits from these stories and whose perspectives are left out.

Through AIxDesign, Piet and her collaborators work to “break the spell” of dominant AI narratives by creating spaces for more plural, grounded, and critical conversations. Their approach combines research, design, community work, workshops, publications, and cultural programming.

Piet presented Better Images of AI, a project developed in collaboration with the organisation of the same name. The project responds to the repetitive and misleading visual language often used to represent AI: blue brains, humanoid robots, glowing code, and abstract digital networks.

These images, Piet explained, make AI appear sterile, mysterious, and detached from the real world. In contrast, the project proposes alternative visuals that reveal the human labour, natural resources, infrastructures, and social systems behind AI.

Images of data centres, mineral extraction, click-work, cloud infrastructure, and AI fatigue help make visible what is often hidden. They remind us that AI is not magic: it is built through material, environmental, and human processes.

Another project discussed during the talk was Feminist UX of AI, which examines how interface and interaction design influence user behaviour.

Piet argued that algorithmic systems do not only act through invisible back-end processes. Their interfaces also shape what users see, choose, understand, and accept. This means that UI and UX design can become powerful sites for intervention.

By reimagining familiar platforms such as Uber, Netflix, and Spotify through feminist human-computer interaction principles, the project explores values such as participation, inclusion, disclosure, and accountability.

For Piet, speculative design plays an important role in making hidden assumptions visible. Often, people accept a platform or tool as “just the way it is” until they encounter another possible version.

As she explained, “sometimes you don’t know there’s an assumption or you think something is just the way it is. And then it isn’t until you see it another way that you realize that other things are possible.”

By showing alternatives, designers can help people recognise that current systems are not fixed. They are designed choices, and therefore they can be redesigned.

Piet also introduced Slow AI, a research direction that challenges the idea that artificial intelligence must always become bigger, faster, and more universal.

Instead of accepting the dominant logic of scale and optimisation, Slow AI asks what could happen if AI were smaller, slower, more situated, and more connected to specific communities and contexts.

This approach opens questions around ecological impact, governance, infrastructure, and care. It also challenges the belief that progress must always mean acceleration.

Within the Slow AI research, Piet explored the relationship between AI and magic. She noted that big tech often uses magical language to make AI seem mysterious or beyond human accountability.

At the same time, she invited the audience to look at older predictive practices, such as rituals and forms of collective interpretation, as ways of thinking differently about uncertainty.

Rather than presenting AI outputs as final answers, Piet suggested that technologies could be designed to foreground uncertainty and encourage users to become active sense-makers.

A major part of the talk focused on AI literacy. Piet argued that knowing how to use AI tools is important, but it is only one dimension of literacy.

Prompting, context engineering, and AI-assisted making are useful skills, but they should not be mistaken for a complete understanding of AI. For Piet, true AI literacy must also include technical, cultural, political, and practical awareness.

It means asking: What is AI? How is it made? Who makes decisions in the process? What values are embedded in it? What alternatives could communities build for themselves?

Piet reminded the audience that AI is not a single technology. It is an umbrella term that includes many different systems and capabilities, from large language models to object detection and computer vision.

Understanding this distinction is essential. It helps demystify AI and makes it easier to recognise both its possibilities and its limits.

Large language models, for example, may feel conversational, but they operate through statistical prediction. They do not understand the world in the way humans do. Recognising this can help users remain critical, especially when facing hallucinations, over-trust, or emotional attachment to AI systems.

Piet also reflected on how AI tools affect creative and cognitive processes. The question is not only whether AI can produce something for us, but how our thinking changes when we use it.

She raised concerns around cognitive outsourcing: the possibility that users may gradually surrender parts of their thinking, decision-making, or creative development to AI systems.

At the same time, Piet did not frame this as a simple rejection of AI. Instead, she proposed mindful adoption: using tools while remaining attentive to personal agency, attention, learning, and creative sovereignty.

Another layer of AI literacy involves recognising AI as a story. Piet described AI as a narrative shaped by companies, investors, media, and cultural expectations.

The language of inevitability, disruption, salvation, or doom can make AI seem larger than life. But these stories are not neutral. They often serve economic and political interests.

To understand AI fully, Piet argued, we must look beyond the tool itself and examine the wider ecosystem: data, labour, infrastructure, materials, environmental impact, governance, and the values embedded across the pipeline.

Piet’s final dimension of AI literacy focused on the possibility of building alternatives. Rather than accepting only the tools offered by major platforms, communities can experiment with local models, open-source systems, custom datasets, and independent infrastructures.

She shared examples of small-scale and situated AI projects, from locally run models to artistic experiments with subjective computer vision and solar-powered language models.

These projects show that AI does not have to be universal, extractive, or centralised. It can also be local, reflective, experimental, and shaped by the needs of specific communities.

In the final part of the talk, Piet addressed the polarisation of AI discourse. Too often, conversations about AI fall into two opposing positions: hype or hate.

The hype position celebrates AI as exciting and inevitable. The hate position rejects it as harmful and exploitative. Piet recognised the validity of concerns on both sides, but warned that both positions can become passive if they only react to what is already happening.

Instead, she encouraged the audience to stay with the complexity of the “messy in-between”. As she put it, “don’t pick a side, even though it’s weird, like being in the in-between, it’s hard.”

This is where people can ask better questions, make informed choices, and participate in shaping more responsible futures.

For the Domus Academy community, Piet’s talk offered an important reflection on the role of designers today.

Designers are not only users of AI tools. They are also interpreters, critics, storytellers, system-builders, and cultural agents. They can challenge dominant narratives, reveal hidden infrastructures, design more transparent interactions, and imagine alternatives.

In this sense, disrupting patterns means refusing to accept AI as something already decided. It means recognising that technology is shaped by people — and that more people should be involved in shaping it.

Piet closed with a call for collective imagination and responsibility. No one knows exactly where AI is going, but its future should not be left to a small group of powerful companies or individuals.

Quoting Paul Eluard, she reminded the audience that “other worlds are possible, but they’re in this one.”

Through AIxDesign and her wider research practice, Piet advocates for collective sense-making, critical literacy, and creative experimentation. Her message was clear: the task is not only to dismantle harmful systems, but also to imagine and craft the worlds we cannot live without.

At Domus Academy, the Disrupting Patterns talk became an invitation to rethink AI beyond automation and efficiency, and to approach it instead as a field of cultural, ethical, and creative possibility.