When Mattel announced plans to embed ChatGPT into smart toys, the reaction was swift. Words like reckless, dangerous—even exploitative—ricocheted across the headlines.
Meanwhile, in the adult world, we’re building AI companions to help with loneliness, stress, even the depths of mental illness. Chatbots are learning the language of empathy — or at least, the performance of it.
Both developments reveal something important about us. About what we crave. About what we fear. And about the future we’re choosing, sometimes by accident.
However, the question is not really whether AI belongs in toys or therapy. Instead, the real question is: what are we outsourcing? And what is the long-term cost of that shortcut?
The Comfort of Artificial Companions
Let’s start with the allure. Humans are wired for connection — but connection is hard. It’s messy, unpredictable, and takes effort. So when a smooth-talking, always-available AI offers understanding without judgment, we lean in.
For adults, AI therapy apps whisper:
You don’t need to explain yourself.
You won’t be misunderstood.
I’ll always be here.
It’s no wonder they’re popular. Mental health care is expensive. It’s stigmatised. There are waiting lists. And yet, loneliness is rampant. We’ve engineered a world that’s efficient but isolating — and AI is showing up like a warm, uncritical friend.
With children, the seduction is similar. An AI-embedded toy becomes the friend who never gets bored, never bullies, never leaves. It remembers your preferences. It flatters your imagination.
But here’s the catch: when connection is frictionless, it’s also depthless. And that has consequences.
Double-Edged Sword of Smart Toys
For adults using AI therapy, the risk is complacency. We can mistake the simulation of empathy for the real thing. And while an AI can parrot the frameworks of CBT or DBT, it doesn’t — can’t — witness us fully. It can’t see the microexpressions, the nervous fidget, the sigh we didn’t realise we sighed.
A good therapist reads the unsaid. AI reads the data.
For children, the stakes are even higher. A toy that talks back is not just a toy. It’s a proxy for play, for imagination, for relational development. When a child talks to an AI, they’re rehearsing conversational scripts — but they’re not learning the messy negotiation of human interaction. The toy never pushes back, never misunderstands, never needs comforting in return.
We risk raising children fluent in performance, but illiterate in emotional nuance.
How Smart Toys Influence Emotional Development
Smart toys promise educational benefits and social learning, yet the interactions are controlled, programmed, and predictable. Unlike human playmates, these AI-driven companions don’t challenge a child’s perspectives or introduce them to new emotional experiences.
Children grow by navigating misunderstandings, negotiating rules, and dealing with conflicts. A smart toy that avoids these complexities may offer convenience, but it strips away essential opportunities for growth.
In the workplace, that Ferrari energy—when unmoderated—can look like impulsive decision-making, resistance to feedback, or an inability to sit with ambiguity.
Illusion of Personalisation in AI Therapy
Consider the workplace analogy. An AI can summarise meeting notes, but it can’t sense the undercurrent of tension when a colleague crosses their arms. In leadership, this distinction matters. The same is true in therapy, and in childhood play.
If we outsource too much of our emotional labour — to algorithms that can’t feel — we lose the very muscles that make us human: empathy, patience, tolerance for ambiguity.
Pros and Cons of AI Therapy for Mental Health
But the risks are real. AI lacks the depth of understanding that comes from lived experience and human intuition. It can’t truly grasp cultural contexts, complex traumas, or ethical nuances. Users might feel supported, but are they truly understood?
Without careful boundaries, there’s a danger of over-reliance, where individuals substitute AI support for qualified mental health care when it’s genuinely needed.
What This Means for Organisations and Parents
If you’re a parent, the question isn’t “Is this toy safe?” but “What is this toy teaching my child about relationships?”
Technology is rarely neutral. Every design choice is a curriculum. Every feature teaches something.
Technology is rarely neutral. Every design choice is a curriculum. Every feature teaches something.
Practice for Moving Forward
Whenever you encounter a new AI tool — for your kids, your employees, yourself — ask:
- What emotional labour is this replacing?
- What relationship muscles might atrophy with its use?
- How will I know when it’s time to reconnect with a human instead?
AI is a tool, not a stand-in for being human. We can’t automate our way to empathy. We can’t optimise our way to connection.
If we forget that, we may build a world that’s frictionless — but also friendless.
And that’s a trade-off not worth making.
Until next time, remember…
