A toy offers a first way to imagine how the world works—and how it might work differently. In this part of my 2024 research into AI and visual culture, I explore toys that could change their responses, appearance or stories through play. The question is not simply how much intelligence fits inside an object, but what room that object leaves for a child's imagination.
AI could help designers explore characters, patterns and forms drawn from a wider visual repertoire. Some toys might change a screen image or projected surface as a story unfolds. Physical changes would still depend on materials and mechanisms. A varied collection of references would not automatically produce cultural diversity: the choices of designers and the people represented would remain important.
Personalisation could begin with a child's explicit choice of a character, language or level of difficulty. A toy might offer a different challenge when a familiar one loses its appeal. This need not require constant recording or a claim to know how a child feels. Parents and children should be able to understand the interaction, stop it and play without an account or an ongoing conversation with a system.
Augmented reality could add an imagined landscape to a room or a story to a physical model. A child might move a real object and see its fictional world respond. AI could help vary that response, but AR and AI would play different roles. The result would be worthwhile if the digital layer enriched physical play rather than making everything outside the screen seem incomplete.
Learning might emerge from asking questions, trying an idea and discovering an unexpected result. An adaptive toy could support that process, but engagement alone would not demonstrate educational or emotional benefit. Not every moment of play needs a lesson. An object that can be misunderstood, repurposed or given a completely different story may offer more freedom than one that always directs the next step.
Custom manufacturing could make different sizes, controls or tactile features easier to explore. AI-assisted design, 3D printing and other production methods might support such variation. Smaller batches would not automatically mean less waste or a lower environmental impact: materials, durability, repair and electronics would all matter. Cultural representation would likewise need considered choices, not an assumption that a system can guarantee respect or accuracy.
I imagine the most interesting toys offering possibilities rather than prescribed performances. They could draw on a shared archive of visual culture while leaving children free to invent their own meanings. A toy's achievement would not be to become the most convincing artificial companion, but to give play more forms without taking authorship away from the person playing.

AI & Visual Culture: 38. Video Games ↗

