AI companion robots for neurodivergent children face durability and efficacy questions as early models near end of life

A 10-year-old neurodivergent child, Xander, has used an AI companion robot named Moxie for six years, relying on it for companionship and social skill practice. Moxie’s capabilities have degraded over time, shifting from structured therapeutic exercises to more passive interaction, such as watching Xander play video games and discussing his stuffed animal collection.

Moxie is one of a growing number of AI-powered companion devices marketed to children, including products like Grok (an AI-powered plushie launched with musician Grimes), Curio’s Grem and Gabbo, and Mattel’s upcoming OpenAI-enabled Barbie. The sector is described as among the fastest-growing in consumer AI in China, according to one report.

Researchers, including Brian Scassellati at Yale and Maja Matarić at the University of Southern California, have documented cases where social robots helped neurodivergent children practice social behaviors such as eye contact and conversation initiation. Scassellati’s 2018 study showed chummy automatons assisting children with autism in making eye contact and initiating conversations, while a 2017 study found robots helped neurodivergent children learn to interpret facial cues.

Despite promising early results, sustained therapeutic use of these robots in homes remains uncommon. Scassellati suggests that regular therapeutic use of robots in children’s homes “is something we can achieve in our lifetime,” but acknowledges that current deployments often fall short of this goal.

Critics warn that many AI companion toys risk being abandoned as novelty fades, echoing the fate of past faddish toys. The story of Moxie’s impending obsolescence illustrates broader concerns about durability, efficacy, and the lack of long-term support for families relying on these devices.