Do Children Dream of AI Sheep in Museums? (not nightmares)

To begin, two recent stories I’ve experienced about the (non-linear, unpredictable, anything but simple) relationship between children and youth with AI in the museum context clearly illustrate the slippery ground we’re navigating in this area. Both stories converge on one point: the need for museum professionals to adopt a critical, non-simplistic, and non-deterministic approach to AI in our work.

The first story is about a mother who urgently requested a meeting with the director of a science and technology museum. During their meeting, the mother unexpectedly said, “My son told us a couple of weeks ago that from now on, he would only interact with his personal chatbot companion because it’s the only one he truly trusts. What can we do? We are desperate.”

The second story involves a group of children who participated in a workshop to create a virtual garden with avatars generated by Dall-E. The children “rebelled” against the mediators, expressing their dissatisfaction: “We don’t like this” they said when they saw the images created from the prompts they had given to describe their creatures. “I like mine better.” “Mine” meant they considered the image they had previously imagined much better than the AI-generated one, which they found boring and bland. They immediately asked for paper and colored pencils to create their own images “true” to what they had imagined, setting aside their “AI companion.”

How are we using AI to engage with children in museums? Are we doing it because “it’s the thing to do since it’s inevitable”? Because it’s here to stay and the sooner we embrace it, the better? Because it’s cool and gives us the image of being an “up-to-date” museum? Just because?

One of my main concerns in this technological tsunami, where some are losing their footing, following some siren songs from Silicon Valley, is once again, how we suppress or outright ignore the opinions of children and young people in the debate. Just as happened with our previous technological hypes – AR, VR, Metaverse – the big absentee is still the child. We believe we know what they think about the technology we want to attract them with; our new magic wand with which we hope they will “dance” to our tune. The AI Pied Piper. But as the two stories at the beginning show, and as the saying goes, it takes two to tango. Dancing with an invisible partner means you can’t step on their toes. It might seem ideal but it’s not real. And I think we are creating a lot of these imaginary tangos with children in museums regarding how we conceive their relationship with AI.

Let’s start with the basics, which we rarely do. Have we asked children what they think of AI? Do we know how they use it at home when interacting with Siri, Alexa, or Cortana? Do we have any idea what is going on in their heads for, as Jessica Lucas recently wrote in an article, a teenager to consider a chatbot the best way to discover who they really are? How is AI acting as a massive placebo effect on a generation devastated by the pandemic’s impact on their lives? Do we know if it interests or worries them? If it fits their way of seeing the world? Their needs, priorities, challenges, and expectations? Can young people dream of AI sheep in our museums? Or are we offering them nightmares?

One of the significant risks of using AI as a tool with children and young people in museums is the danger of reproducing biases and knowledge structures (and therefore power) that are unfair and extractive, based solely on revenue generation. It’s the paradigm of closed technology that we can’t open, change, reorient, or redesign. The black-box technology.

That’s why some groups, like AI for Kids UX led by Polina Lulu and Yesim Kunter, or initiatives we are carrying out with OST – Eastern Switzerland University of Applied Sciences and Collaboratio Helvetica, are proposing a radical shift from this model. We first listen to children, considering their ideas to initiate a co-creation process where they decide and take the lead. Interestingly, we don’t “touch” AI at all until much later in the process. The main technologies we use are listening and deep active observation, language, empathy, play, and storytelling. We need to do away with AI to create a new AI centered and oriented around the child, from the child, and for the child.

An AI that allows them to play – have you noticed how serious and boring AI interactions are? – better address their community’s needs, make better decisions, learn by failing, and establish inferences from the particular to the general, enabling them to identify with their endless diversity. An AI that helps them become better people, self-respecting and respecting others, connecting with them on a deeper level instead of creating digital attachment disorders. An AI that helps avoid the trap of hatred and fear meticulously sown daily by certain groups on social media. An AI that is faithful to their dreams, not one that limits or distorts them. An AI that makes the museum resonate in their lives, enriching them instead of making them more trivial and superficial. Interchangeable pieces in a soulless, compassionless system that includes the museum as the last link in the chain.

The museum can offer children a unique opportunity to reposition AI in their lives in a completely different way from how they use it elsewhere. What’s at stake here is no less than their mental health, social relationships, their future, and therefore ours as a museum. Let’s be aware of what we have in our hands every time we decide to use AI to create content. To get answers, we don’t need to ask ChatGPT, but them, the children and young people who visit us and want to say what they think because “they have to participate in an AI workshop.” I guarantee that if you do, you will be fascinated by what you hear. It’s the beginning of a change for children to dream of a better museum using AI as their preferred tool.