One of the great powers of modern technology is its ability to make us forget what things are for. As social theorist Jacques Ellul writes in The Technological Society, “Our civilization is first and foremost a civilization of means; in the reality of modern life, the means, it would seem, are more important than the ends.” While written in 1954—well before the advent of most modern digital technologies—Ellul’s observation holds purchase today. In the health-care sector, for example, advancements in electronic medical record-keeping (i.e., gains in efficiency in how we store and transfer medical information) often distract from, if not get in the way of, patient care. This type of reversal often narrows our original goal, as gains in efficiency almost always come at the cost of a more holistic vision of the world. 

Welcome to AI-Powered Childhood

Consider the rapid emergence of “smart connected toys,” what I will call for short SCTs. The development of SCTs is not surprising—what object in the world won’t somehow change with AI?—but it is an important case demonstrating the ways we should discern the appropriateness of new and different technologies in our life. 

According to social scientist Jochen Peter and his team of researchers, SCTs are defined by a constellation of six attributes: They are powered by energy (e.g., a battery), rely on sensors, are software-controlled, are interactive (i.e., they can react to inputs and are not static), possess mobility, and are embodied rather than virtual. The Furby, first released in the late 1990s, might be said to be the first SCT: a mass-marketable toy that enabled kids to “interact” with its various sensors (blinking eyes, wiggling ears), to say nothing of the “furbish” language that kids picked up. 

As artificial intelligence increasingly makes its way into more arenas of modern society, it has inevitably made its appearance in the world of SCTs. The AI toy industry, which includes embedding chatbots into characters, robots, and teddy bears, is projected to reach $106 billion within the next decade. Like much toy manufacturing, this trend has accelerated the fastest in China, where over fifteen hundred AI toy companies were operating by 2025. The AI toy market is reportedly growing faster than any other branch of consumer AI. Consider BubblePal, an AI-powered necklace that can be draped over a child’s favourite stuffed animal, enabling “smart” conversation with the previously mute (dumb?) teddy bear. “BubblePal is your ultimate source of knowledge,” marketers promise. “Ready to answer all your questions and become a ‘super encyclopedia’ at your fingertips.” Since its release in July 2024, 250,000 units of BubblePal have been sold. Most recently, Mattel has announced a future partnership with OpenAI, with the goal of embedding AI-powered technology in a series of new Barbie dolls (as if Hello Barbie, released in 2015, had not raised enough privacy concerns). Along with “generic” toys like BubblePal and future Barbies, plenty of SCTs are marketed toward specific activities: Moxie helps with “social emotional learning.” Miko teaches STEAM skills. Snorble assists with getting kids to sleep. 

The Appeal and Danger of Smart Connected Toys

The appeal of SCTs is obvious: They are interactive and personalized. Some can interface with existing toys. They are marketed to encourage imaginative play. They promise to “solve” various challenges of parenting (what parent has not dreamed of a robot to get their kids to fall asleep faster?). And for parents everywhere who are sick of their kid staring at a glowing rectangular box, they are happily “screen-free.” Marketing agents from Snorble sum it up: “Our smart companion grows with your child by combining the best in tech, education, and healthy habits (such as quality sleep) to give every child a chance to excel.” 

As appealing as these features of SCTs might be, I do not think it takes an expert in adolescent development to intuit that they are also profoundly disturbing. But if you are looking for such expertise, consider the recent remarks by Mitchell Prinstein, chief of psychology for the American Psychological Association, before the US Senate Judiciary Committee, on the harm of AI chatbots. Prinstein discusses and identifies many of the “typical” critiques of AI in the context of adolescent development: Algorithmic bias perpetuates various social stereotypes (SCTs marketed toward girls will talk about clothes and toys, while SCTs marketed toward boys will talk about lasers and science). Few AI toymakers have remotely considered what should be done with the data that children share with their SCTs, including the possibility that a child will disclose information about disturbing home environments (a child may trust an SCT more than their parents or other adults, given the “sycophantic” nature of all AI chatbots). Investigations have determined that some chatbots, most notably Character.ai, have even encouraged the violent thoughts of users, including violence between children and their parents. The obsequious nature of SCTs erodes the possibility of forming pro-social coping and conflict-mediation skills (human relationships are never so frictionless). And finally, the activity of children’s play becomes one more arena of human life to commodify, transforming something of intrinsic worth into something that can make a profit. SCTs are, in short, largely not oriented toward the flourishing of the children with whom they interact. 

Interrogating the World of Smart Connected Toys

Discernment of an emergent technology’s appropriateness tends to operate on two levels. The first is direct and particular: What are the immediate harms or benefits associated with the device? A ledger is drawn in which the “good” is placed on one side and the “bad” on another—if the scales tip toward good, then the technology is good and should be adopted. Prinstein’s remarks operate on this level, as does much of modern public health. Within this framing, the SCT itself is neutral. Sure, there are data-privacy concerns, but perhaps the “personalized learning” that SCTs enable makes up for that. While algorithmic bias may perpetuate stereotypes, at least SCTs give kids an opportunity to not look at one more screen. And so the calculation goes. 

The second level is indirect and more general, but also more important. It asks: What kind of world led to the creation of this technology—basic assumptions, desires, visions of the good life, and so on—and is that the kind of world we wish to live in? If the first looks directly at the device, as if through a microscope, the second positions the device itself as the lens through which the surrounding world is revealed. Within this framing, technology is never neutral, SCTs included. As Pope Francis famously states in Laudato Si’, “We have to accept that technological products are not neutral, for they create a framework which ends up conditioning lifestyles and shaping social possibilities along the lines dictated by the interests of certain powerful groups. Decisions which may seem purely instrumental are in reality decisions about the kind of society we want to build.”

In pursuing this second form of analysis, we might ask, following Francis, What kind of society do the makers of SCTs want to build? Combined with our earlier observation from Ellul, we might amend this question to something like, What kind of purposes do the makers of SCTs want to blind us to? 

Rediscovering the Purpose of Childhood

There are many good answers to that question, but I want to explore just one: what a world built around the development of SCTs reveals about what childhood is ultimately for, especially the role of parents in that world. 

I take it that a child’s imagination is one of the most precious resources a society can protect, and this is especially the case in the twenty-first century. Any adult who steps back for a moment to observe children at play should stand in awe. The combination of the limited resources required with the magnificent worlds that are built in their minds is astonishing. Carboard boxes, sticks, a few dolls can yield complex narratives of adventure, rich settings, and long character-development arcs. When children are at play, they are doing something natural that most adults have forgotten how to do. When was the last time your imagination was stretched to such impressive heights? 

When a digital interface is introduced into this activity, there is inevitably a kind of flattening-by-replacement. As psychologist Peter Gray observes, one of the most beneficial aspects of free play in the real world is that kids must act as legislators and juries in the interactive and reflexive encounter of the space around them. In most video games, these roles are fulfilled by the platform. Something similar could be said for a child’s interaction with most SCTs: The kind of interaction they provide is always, in a way, standing in for the kind of interaction that children would otherwise enact themselves (especially between multiple children), and therefore robs them of all the “secondary” goods associated with having to enact that interaction themselves (e.g., pro-social coping and conflict mediation skills). 

When children become used to finding answers to all their questions (why this? why that?) from, for example, the “super encyclopedia” of BubblePal, they are conditioned to receive them as the “definitive” answers to their questions. But the reason children delight in asking questions is not because they are desperate for ready-made answers to solve some abstract problem they’re working through (case in point: their next question is entirely unrelated to their first question), but because they are full of wonder at the world, and delight in the very asking of the question. Whether in the context of frivolous questions or frivolous games, SCTs flatten this activity in a direction that diminishes the power of the imagination. If childhood is in part for enriching the power of a child’s imagination, and not just for accumulating information, it is hard to see what productive role SCTs can play in that purpose. 

While tragic, I don’t think this flattening represents the worst impact of an SCT-powered childhood. 

To appreciate why childhood imagination is such a precious resource, one must grapple with the absence of imaginative power in the modern world. Neo-Marxist social theorist Theodor Adorno, writing in the 1960s, was especially critical of how a society built on unconstrained economic productivity and capital accumulation tends to destroy our potential for imagination, given that the kind of work many of us engage in is not the kind of work that demands much of our imagination. Mindless jobs lead to mindless recreation and back again (or so the Marxist critique goes). While his judgment is sweeping, I think this pattern can be seen, to varying degrees, in many forms of modern work and recreation. Because of this, Adorno observes, we even struggle to conceptualize imaginative uses of our “free time” (a category Adorno thinks is itself indicative of our problem—a small slice of time in which we are finally “free”): “The impertinent question of what people should do with the vast amount of free time now at their disposal . . . is based upon this very unimaginativeness.” This is not the place for a full-blown review of Adorno’s thought or Marxist analysis, but a casual glance at the number of adults who habitually binge-watch Netflix and other streaming services lends significant support to his claim. The widespread experience of boredom suggests that most adults are desperate for some source of wonder. When was the last time you observed a line of adults—waiting for check-out, waiting for the bus—not glued to their phones? When was the last time you asked a “why” question the likes of which children love? 

While Adorno would have never suggested this—consumed, as Marxist are wont, with purely material solutions to material problems—I think one purpose of childhood might be to save us adults from the soul-sucking numbness of an otherwise wholly unimaginative world. 

Consider, first, a similar deployment of AI for the sake of reading proficiency. Ying Xu, an assistant professor at the Harvard Graduate School of Education, has considered the role of “AI companions” to help students learn how to read. These companions can read the story aloud and pause to ask questions to the child, therefore engaging the child, according to Xu, more “actively” in the reading process. Xu concludes with a shocking parallel: “And the AI listens to the child’s response and offers small hints, just as a caregiver or a teacher would if the child needs help.” 

Just as a caregiver would. 

The stubborn fact of the matter is that when AI reads stories to children, a caregiver is not there. Therefore, that caregiver never enters the imaginative space of their children’s story. They never receive the myriad “whys” that gush from the mouths of little kids and therefore must never address those “whys” for themselves or pause to consider the importance of asking questions that begin with “why” as opposed to those that begin with “how much.” C.S. Lewis writes in the dedication to The Lion, the Witch and the Wardrobe, “Some day you will be old enough to start reading fairy tales again.” While dedicated to his goddaughter Lucy, Lewis’s admonition is really directed at all of us who have as much to gain from the reading of fairy stories—and stretching our imaginations to do so—as our children do. Contra Xu’s technological optimism (or naïveté), replacing the role of the adult with an AI chatbot does have an effect, starting with the adult themselves. This is, at least, how I have experienced my own time spent as a parent reading to my child. Every time I read a new story to my daughter, my soul is lightened, especially when the events of the day have been draining. I leave it to the parents who read this article to consider whether the same is true for them. 

What is true about reading fairy tales is broadly true about all the ways parents and other caring adults engage with children in imaginative play. For an adult to help their kid build a fort, dress up as a fanciful character, participate in a silly made-up game (whose rules change as often as they’re broken), or jump into a spontaneous dance-a-thon in their kid’s bedroom is to step out from what Josef Pieper once described as the “workaday world,” where everything is measured by efficiency and economic productivity—the kind of world Adorno so strongly critiqued. To replace the role of the parent in this unique form of creativity is, in no small way, to destroy one of the few remaining powers that can bring life to an otherwise disenchanted world. How can we not defend (and personally adopt) the particular virtues of childhood from the alienating forces of the machine? 

I don’t want to be naive toward the legitimate challenges of parenting that make finding time for this kind of engagement difficult, and that therefore make the appeals of SCTs even more powerful. This is especially the case for working parents who have limited time to spend with their children and are looking for sources of reprieve. Learning from the imaginative capacity of one’s child is not going to solve this problem. (Adorno’s critique of modern imagination is, of course, steeped in a much fuller critique of modern working conditions.) However, for moments in our day in which we do have the freedom to decide how to spend time with our children, we can enjoy this aspect of childhood more than the marketers of SCTs imagine is possible. To the extent we see SCTs as a supplement to this activity (the marketed goal of SCTs), we risk SCTs becoming a replacement of this activity (the ultimate goal of SCTs). 

The meaningful engagement of a parent in the imaginative life of their child is also—contrary to all the “best in tech, education, and healthy habits” that SCT marketers spin up—an opportunity for our children to step out of the workaday world. I cannot think of a single image more opposed to our hyper-efficient world of economic productivity than a child left to their own imagination and at play. And yet, in response to the question “What is childhood for?” it would seem as though the builders of SCTs presume the answer is something like “Learning facts about the world as efficiently as possible.” What a four-year-old needs is manifestly not an “ultimate source of knowledge,” as if having all the information on the internet at our fingertips had proved to make us better people, but rather a personal source of limited knowledge—their parent. Indeed, if one scratches below the surface, the real “Anxious Generation” is not just those kids who grew up with the iPhone but also their parents, who felt it necessary to give their ten-year-old all the tools they needed to “excel” within the workaday world. Is childhood really about achieving “excellence”? 

It is notable that the inability of SCTs to provide imaginative moments between parents and their children would remain even if there were not so many issues associated with data privacy, or if the likelihood of AI chatbots sharing harmful content was not so high, or if the kind of personalized learning that SCTs provided was more “efficient” than some befuddled adult could provide. You do not need to be the chief of psychology to appreciate how SCTs undermine the possibility of flourishing parent-child relationships. You only need to have the courage and patience to look at the children right in front of you and attend to how your interactions with them change your life. As Ellul warned, a strict analysis of the immediate pros and cons of a specific technology often distracts us from bigger questions about purposes that the technology tends to crowd out. 

What the introduction of SCTs should do, therefore, is what the introduction of all AI-powered devices should do: make us step back and ask, once again and with great courage, what our most basic human connections are for.