Categories
reflections

New relations in generative things

On 4 June 2024, I gave a talk at an evening on “Designing Intelligent Cultures with Data and AI”, organized by design agency CLEVER°FRANKE from Utrecht (check also their recap). I decided to share my developing thoughts on the emergence of Generative Things, a new breed of things that we foresee happening in a certain form in the research of Cities of Things for some years, but with the rise of the new generative AI could become even more relevant to explore. Especially for designers. My goal for the talk was to give a kind of direction of thinking that -in my humble opinion- is needed to design these new things, or better said, the new relations with these generative things.

As background for a transcript of the talk, I recorded a test run that I did as practice (and timing), and I feed this together with the slide deck to ChatGPT. I then asked to rewrite the first rough breakdown of slides and transcript into a blog post. It did a good job making a summary, and I did the final edit afterward. Below is the result. Let’s start by sharing the slides. If you like to know more, get in contact.

“Today, I’d like to take you on a journey exploring new relations with generative things. My main goal is to make sense of the unpredictable futures of human-AI partnerships. This journey is reflected in my activities and research, which you will see in this presentation.

No Future Without a Past

There is no future without the past. The generative things we talk about today may be new, but they have roots in concepts we’ve been exploring for years. For instance, we once looked at smart objects like the Nest thermostat as early examples of AI using pattern recognition to predict future uses.

This presentation revolves around three key questions:

1. What will generative things be like?

2. How will we understand the intentions of these new co-performing generative things?

3. What will this mean for designers of generative things?

Understanding Co-Performance with AI

I started my research journey with the “Cities of Things” initiative at Delft University of Technology in 2017. We explored the concept of smart cities, not just in terms of human behavior data collection but also considering intelligent objects with agency. We asked, “What if things were like citizens in our cities?”

In our 2018 paper, we used design fiction to explore a future where humans and agentic things co-exist. We identified several dilemmas:

  • Responsibility: Who is responsible, the private entity or the public?
  • Priority: Who takes priority, humans or the system?
  • Relationship: Are these things tools or social partners?
  • Adaptation: Should adaptation be more human-centric or thing-centric?

Co-performance, a concept from Lenneke Kuijer and Elisa Giaccardi, inspired us. It involves shared goals between humans and intelligent technologies, working in concert. This idea of shared goals and collaboration is crucial in designing future interactions.

Exploring Urban Robots

One of our projects, “Hoodbot,” in collaboration with the Rotterdam University of Applied Sciences and funded by the Municipality of Rotterdam, focused on designing urban robots that are not introduced in cities and neighborhoods by big tech companies, but are designed and created by the residents themselves. We engaged citizens to co-design prototypes, understanding how these robots could become social partners rather than mere tools. This project highlighted the importance of robots’ social roles, like facilitating community interactions in a garden. This student team project started as a robot that would help you with gardening but evolved into a task orchestrating and connecting spider in the web of the community garden.

Living in a Digital First Reality

We are already living in a digital-first reality where AI will become fully immersed into our lives as next phase. The breakthrough moment for AI was on November 30, 2022, with the introduction of ChatGPT. This interface made AI interactions more accessible and opened new applications for human-AI collaboration. Not in the least, as it became easy to imagine what you can use it for. It has become a companion and an agent in our everyday lives. With the latest version of GPT-4o, the promise of a combination of natural communication and interaction forms that function without latency makes it possible to pass the uncanny valley for AI-bots.

Generative Things: The Next Iteration?

We are now exploring if generative things could be the next iteration of AI. Early attempts by companies to create multimodal AI devices, such as smart glasses (Meta), pins (Humane) to wear or dedicated companion devices (Rabbit) show promise but also face challenges like latency and battery life. The essence of these devices I think, is how they bring AI closer to us, creating an even more integrated interaction. AI becomes an always-present companion.

The Generative Things go even beyond these AI devices. Mundane things we use in our everyday lives will be enhanced with intelligence and conversational relationships, which might lead to different relations with them.

Addressing the Chilling Effect

We must consider the “chilling effect” of synthetic reality, where perceived observation alters our behavior. As things with agency become part of our ecosystem, we might start to adapt to their intelligence and behaviors—just like we start adapting our behavior in a street fully equipped with smart doorbells. More interesting than discussing privacy concerns is what it does to the neighborhood community, how it influences how people respond to each other, losing a safe space. It can create tension in communities.

The Need for a New Language

To navigate our interactions with generative things, we need a new language. This language should communicate the learnings and intentions of AI, enabling continuous reinforcement learning from human feedback. Transparency and especially contestability are key principles, allowing us to scrutinize AI decisions and interactions.

Principles for Human-AI Interaction

Ethan Mollick’s principles for AI interaction as described in his book on co-intelligence offer a potential framework:

1. Always invite AI to the table. “(…) familiarizing yourself with AI’s capabilities allows you to better understand how it can assist you—or threaten you and your job.”

2. Be the human in the loop. Collaboration creates better outcomes, and we should be the guardrails.

3. Treat AI like a person (but define what kind of person it is).

4. Assume this is the worst AI you will ever use.

Designing for Unknown Unknowns

Finally, we are now in a phase of designing for unknown unknowns. We don’t fully understand how generative things work or think, but we must create a framework to deal with them. This requires innovative thinking from designers of data, visualization, and interaction.

Thank You!

I hope this talk inspires you. You can follow my writings weekly, where I explore new aspects of context and generative things. We are also planning to make generative things the theme for the next ThingsCon event. Let’s continue this discussion and explore how these generative things will shape our future.