Once upon a time, Geppetto created a puppet named Pinocchio. Though he had carved him out of wood, due to the particular way Geppetto had designed him, Pinocchio came alive. Or at least Geppetto thought he came alive, because Pinocchio did all the things a real human boy would do. He walked, he talked, and he played tricks! He also got up to antics, like staying up late reading pirated copies of books on the Internet.
Because Pinocchio was so much like a real boy, Geppetto gave Pinocchio chores to do. Pinocchio was always willing to help, but unfortunately, he had a bit of a problem with the truth. Whenever Geppetto asked Pinocchio to clean his room, Pinocchio would quickly sweep his toys under his bed, claiming he had worked hard to leave his room spotless. He lied so much, in fact, that his wooden nose, at first short and stubby, started to grow longer with every lie.
We’ve done something similar to Geppetto with Artificial Intelligence (AI), specifically those AI models with Machine Learning (ML) capabilities that can generate content (aka generative AI), such as large language models (LLMs). We’ve given these models goals and agency, but unlike Pinocchio, AI doesn’t have a “wooden nose” to show us when it’s misleading us or how exactly it accomplishes the tasks we give it.
This might be harmless when we ask ChatGPT to help us make dinner with the random items in our pantry, but what happens when we use it for advice on our interpersonal relationships, to write legal briefs that inform court rulings, or to aid government decision-making that reshapes the geopolitical landscape?
Bodystorming Artificial Intelligence
In early 2024, we started having conversations about how generative AI was impacting research and human-centered design. These conversations culminated in a series of workshops called Bodystorming Artificial Intelligence, which we ran at EPIC 2024 in Los Angeles and UXLx 2025 in Lisbon.
Photographer: Jose Goulao
Using bodystorming (an ethnographic method in which participants physically act out situations they are trying to innovate within), we set out to predict the unpredictable. We hoped to uncover the specific unintended consequences of using generative AI, and to develop a set of ethical principles that could be applied to designing this rapidly evolving technology.
However, the consequences we discovered were less about AI, and more about how humans might behave and engage with AI in different scenarios. And because AI is evolving faster than humans or ethical frameworks ever could, starting that far downstream felt like we weren’t doing much more than trying to put a genie back in its bottle.
Simply moving upstream to try and shape AI policy for a product or organization wouldn’t be enough either, as policy governs AI use, but does not address the fundamental question of who is included and who is left behind in a human-centered approach. Shifting upstream also fails to account for the most significant impacts of AI, including the increasingly unexpected and far-reaching ways in which AI is changing how humans think, behave, and interact with one another.
We need to zoom out to see the full picture of how AI is transforming us on an individual and societal level.
“Technologies are not neutral. They affect the course of society, aiding some actions, impeding others, independent of the morality or necessity of those actions. Technology also has its side effects, both physical and mental. Technology can aid as much as it can detract. It really is up to us, both as individuals and as a society, to decide which course we shall take.”
We had set out to test ethical principles for AI; instead, we ended up questioning how we as humans relate to and incorporate artificial intelligence into our lives.
AI and Human Agency: A Day in the Life of Geppetto
Let’s imagine a day in the life of Geppetto in 2025. Geppetto, a modern-day creator, has designed Pinocchio, a generative AI program, to keep him company. Pinocchio can do much more than just chat with Geppetto, and quickly seems to take on a mind of his own. So Geppetto starts asking Pinocchio for advice, to help complete daily tasks, and to even come up with ideas for creative projects.
On a typical day, Geppetto might ask Pinocchio to help with a task for work, give ideas on what to eat for lunch, or provide feedback on a problem. Pinocchio might also work behind the scenes in ways that Geppetto doesn’t even see, like filtering and summarizing his search results and shaping his media feeds.
There’s only one problem. Geppetto doesn’t know when Pinocchio is lying because he didn’t design the AI program with a modern-day equivalent of a wooden nose. Plus, Pinocchio is very confident and convincing in how he provides Geppetto with information and advice. But because Pinocchio is so helpful, Geppetto decides it doesn’t matter. Besides, it is Geppetto, not Pinocchio, who has control and agency over his own life.
Or is it? Let’s examine Geppetto’s day more closely.
During a day in the life of Geppetto, Pinocchio has intervened and nudged Geppetto to act or feel in some way no less than 12 times! Pinocchio has not only shaped Geppetto’s work for him, he has also determined the food that goes into his body, influenced his moods, impelled him to justify a decision he did not make, advised him on the type of relationship he should have, and even suggested he create another version of Pinocchio as a replacement for human connection.
Pleasure Island, or Donkeyland
“That happy land of carefree boys, where every day's a holiday!”
In Disney’s Pinocchio, Pleasure Island is a place where children go to have unlimited fun and freedom, rather than attending school and doing chores. Stay too long and the children become donkeys who have sacrificed their humanity and lost their ability to communicate. Are we headed toward a future where we’ve outsourced so much of our human ability that we also devolve into donkeys?
We’ve already given technology the power to shape much of our lives so we don’t have to. Modern life is replete with so many micro and macro decisions that technology serves as a sort of antidote to the overwhelming amount of choice we have created for ourselves—between 20 brands of toothpaste, between dozens of streaming services and thousands more television shows, or the seemingly unlimited number of people to sift through on dating apps.
If a core challenge of modern life is decision fatigue, the embrace and rapid incorporation of AI is understandable. It’s a technology designed in our own image that promises to help us deal with this overwhelming amount of information so we have time and energy for the things that matter most to us.
How Humanizing AI Might Actually Help Us
One thing we learned in our bodystorming sessions was how much we as humans can’t help but anthropomorphize, or humanize, AI. Even without bodystorming, where the whole point of the activity is to roleplay and step into another’s shoes, users of AI tools like ChatGPT are inclined to give it a human name or a more distinct human personality. It’s almost as if we want AI, already designed to mimic human thought and speech patterns, to feel even more human than it already does.
Anthropomorphization is a common way we, as humans, relate to non-human things. This includes living things, such as pets and plants, and non-living things, such as ships (“she’s listing starboard!”) and designer handbags (“isn’t she a beauty?”).
AI is not a living thing, but it’s not quite a static, non-living thing either. Generative AI has agency, to an extent. Tools like ChatGPT have goals and the ability to learn new things independently, which means they change over time. And because these tools can learn and change so rapidly, even technologists can’t always tell how they come to conclusions and make decisions.
Geppetto couldn’t help but think of Pinocchio as a real boy because he acted like one. While there are real dangers to anthropomorphizing AI, our experiences with bodystorming this technology illuminated one positive of humanizing AI—the potential to critically examine our relationship with it.
If we think of generative AI as an independent, evolving entity, we might question how much of our lives we want to cede control over to these entities that constantly intervene and act on our behalf.
If we think of generative AI as Pinocchio (a puppet we designed that became a real boy), our focus shifts from designing ethical principles in an effort to regain control to developing principles for engagement with an independent, unpredictable, living agent.
Which then raises a few questions:
- How might we develop these principles for engagement to be resilient in the face of constant and rapid technological change?
- How might we become more conscious (and conscientious) of how we’re engaging with AI in the first place?
- As researchers and designers, how might we help to make the invisible ways AI influences us and our decisions more visible?
Finding Our Jiminy Cricket
Bodystorming can be a useful and predictive activity when you’re trying to play out what independent agents might do. It helps people think through what is happening beyond their own point of view, and the repercussions of their actions and decisions. It can also help us break free of entrenched thought patterns and approach something like AI with curiosity rather than fear.
The goalposts for AI have shifted, and they will only continue to shift. The questions we need to be asking are not about how we can design AI ethically after the fact, but about how we might engage with AI with greater awareness and intention, making more deliberate choices about how, when, and where we choose to let AI intervene in our lives.
In Disney’s Pinocchio, Jiminy Cricket was Pinocchio’s conscience, a role specifically designed to help guide the puppet toward becoming a real boy. As we think about AI’s rapid development and our engagement with it, consider what our collective Jimmy Cricket might tell us to do:
Would he tell us to lean into our uniquely human abilities—to feel emotions, to pay attention and be present, to commune with other humans, to consider and deliberate, to choose?
After all, what is more human than to consider and to choose?
About the Authors
Jane Park Storm is a Senior Principal Design Researcher at Coforma with 17 years of experience in the nonprofit and government sectors. She’s spent her career designing and executing socially innovative programs and products for positive global and civic impact. Jane has a BA in Philosophy and minor in Creative Writing from U.C. Berkeley, and is the former Director of Product and Research at Creative Commons.
Daniela Jones is a Principal Design Researcher at Coforma with 9 years of experience in strategic design across the public and private sectors. As a researcher and behavioral designer, she blends behavior change research and human-centered design methodologies to deeply understand how people think and behave to advocate for meaningful, evidence-based design solutions. She has a BA in Psychology from the University of Virginia and a Master of Behavioral and Decision Sciences from the University of Pennsylvania.
