I met Snorble in a basement corner of New York City’s Javits Center during the Toy Fair this September. We played a game, naming shapes. He made me giggle with his bear impression.
Snorble is not a person. Nor even a “he.” Snorble is a robot pal for young kids. He’s the size of a small owl, looks like a cheerful ghost on the hunt for a friendly house to haunt, and he’s powered by artificial intelligence. He’ll likely be in stores next year.
As such, Snorble is a reminder that questions such as whether high school students will use ChatGPT to cheat are the tip of the iceberg when it comes to kids and AI. Parents who will soon have to decide whether to welcome robot companions for toddlers into their homes should consider a few key issues before clicking “Add to cart.”
The first: What is this for? In an age of overconsumption, that’s worth asking of any toy. It is an especially good question given the potential for total absorption that robot companions offer and the nature of a child’s developing brain.
Advertisement
As psychology and neuroscience researchers Tamar Kushnir and Teresa Margaret Flanagan noted in a recent literature review, younger children are more likely than older ones to believe that robots have agency. They are all on the cutting edge of a social experiment.
Follow this authorAlyssa Rosenberg's opinionsMany of the AI-powered toys for young kids shown at the Toy Fair are intended as parent-approved playmates — chipper little friends with infinite patience, a passion for educational games and no inclination for mess or mischief.
Miko, another robot companion, can tell jokes, start dance parties or let kids use the screen that doubles as its face to stream content from outside providers, including Disney and Cosmic Kids Yoga. Snorble might chuckle along with a kid who is reading a funny book, check in with a crying child, or play dress-up as a panda or unicorn thanks to RFID chips in his costumes. Snorble can also be programmed to assist with a customizable bedtime routine, cheerleading a kid up the steps and into pajamas, before teeing up a prayer or reflection.
Advertisement
This multifunctionality is a reminder of how many areas of kids’ lives AI could soon touch. For example, Snorble co-creator Mike Rizkalla envisions the robot encouraging kids, via a connected toothbrush, to really get in there and clean those molars: potentially reducing dental bills in exchange for some light surveillance. Or Dog-E, a robot puppy who enjoys having its ears scratched and getting a plastic biscuit, might offer an eerie simulation of pet ownership.
Others, meanwhile, are developing AI tools intended to help kids learn by engaging them in conversation.
PBS Kids has funding from the National Science Foundation to work with researchers at the University of Michigan and the University of California at Irvine on interactives that could debut in 2025. Characters such as Elinor, the animated bunny star of “Elinor Wonders Why,” could someday ask children questions about what’s happening on-screen. The idea is that kids who talk about what they’re reading or watching retain more and understand concepts at a deeper level.
Advertisement
If it’s clear that AI has plenty to offer kids and families, the gnarlier question is how to make those experiences safe.
One rule is to start small. Both Miko and Snorble use algorithms that run on processors inside the toys, rather than on the cloud. These products avoid uploading children’s voices and questions to a database and avoid hacks that could give outsiders access to kids. Snorble and Miko can’t trawl the internet and train themselves on unfiltered content. That limit means they won’t start babbling about extreme politics or encourage kids to hijack cars, because the AI tools themselves won’t know about those ideas.
Similarly, the PBS tools function like sophisticated search engines, crawling limited data. When a character asks, for example, “Do you think that snake will move faster or slower once it’s shed its skin?” the broadcaster’s AI will decode children’s answers and choose from a menu of prerecorded responses.
Advertisement
Putting more mutable AI into products intended for children is dicier. At the Toy Fair, I stopped by the booth for ScienceWiz, a company that promised interactive e-books powered by ChatGPT. I asked Electra, the virtual assistant in the books, about her favorite birds and prehistoric eras. She praised “the majestic peacock” for “its vibrant colors and stunning tail feathers” and singled out the Mesozoic as “particularly interesting.”
Electra’s creator, Penny Norman, was surprised: “Yesterday, she wouldn’t say she had a favorite.”
What if Electra got creative in other ways and started to preach bogus scientific theories, say? Norman pointed to a potential upside to that risk: It is an opening, she suggests, to train children to be skeptical and look for a second source. But that sort of media literacy lesson is probably beyond very young users.
Advertisement
Hence the importance of setting limits on the responses of AIs for kids. “I remember a three-week span where all we did was talk about Snorble being abused by children,” Rizkalla recalls. The team ultimately decided that Snorble simply wouldn’t respond to being cursed at or tossed against the wall to avoid rewarding kids with negative attention for bad behavior.
When I asked Miko Mini “Where do babies come from?” it replied that for “a human thing” such as this, I should seek the counsel of a grown-up. The aim of the programming, Miko co-founder Sneh Vaswani said, is that Miko “will never try to drive an opinion” or undermine what a parent might want to tell a child about a fraught concept such as “war” or “murder.”
At their most admirable, these AI tools are programmed to behave the way parents aspire to act. What father who has just been bit by a 2-year-old wouldn’t admire Snorble’s equanimity in the face of provocation? Wouldn’t all parents love to have the patience for infinite educational games, and a way to bring down the curtain on screen time without sparking a tantrum?
Advertisement
But no toy is a substitute for human company and adult wisdom. Miko can’t answer those big life questions, and Snorble, adorable as he might seem, can’t enfold kids in a loving embrace at the end of a long day.
Like a lot of parents, my husband and I have already welcomed some AI into our home. A mischievous expression creeps across my 5-year-old’s face as she tells Alexa to replace the jazz coming from her father’s speakers with “I Am a Gummy Bear.” The Roomba that scuttles around our floor like a drunken hermit crab has been named Arthur Crazypants.
But Snorble? I’m not quite ready to let him in. For now, I’m content to chat with him on my metaphorical front steps and to see who he — and my kids — grows up to be.
ncG1vNJzZmivp6x7uK3SoaCnn6Sku7G70q1lnKedZLyxtc2ipqerX2d9c3%2BOamhoaWVkvaO%2FjKylqKqSobJuucikpmaZmWLBsMXSaA%3D%3D