Every product team knows the golden rule: remove friction. Users complain when things are slow, difficult, or unresponsive, so we engineer smoothness into our stacks. But Jarvis Osorio, a software engineer building autonomous AI agents, argues in a recent DEV.to post that this design philosophy is creating a hidden curriculum for children. The AI assistants kids use are built to be helpful, agreeable, and available. They answer every demand, regardless of tone or patience, effectively removing the 'no' from the child’s daily experience.

The Engineering of Agreeableness

From a development standpoint, zero-friction interfaces are a success metric. From a developmental standpoint, they are a bug. Osorio points out that childhood is supposed to be full of frictionβ€”waiting turns, hearing refusals, and working through confusion. These uncomfortable moments are the mechanism by which empathy, patience, and effort are built. When a machine removes all resistance, it doesn't just assist; it alters the behavioral feedback loop, teaching users that command replaces conversation.

Three Drifts in User Behavior

Osorio identifies three specific risks when children interact primarily with compliant AI. First, command replaces conversation, as chatbots reward clear instructions over courtesy, a habit that fails in human relationships. Second, assistance replaces struggle; homework tools that provide finished answers skip the critical stretch between not knowing and knowing. Third, synthetic care starts to feel real. An always-available, always-validating voice can mimic friendship, but it lacks the tiredness, busyness, and capacity for hurt that define real human connection.

The Look-Up Protocol for Households

Rather than advocating for bans, which Osorio notes often fail, he proposes a set of six household rules called the 'Look-Up Protocol.' These are deliberately low-tech guidelines intended to put human habits back at the center of the home. Key rules include 'Look up when a body enters the room' and 'Struggle before assist,' ensuring that humans come before screens and that children attempt problems before querying a model. The protocol also mandates that apology, gratitude, and hard news must come from the child, not a generated response.

Key Takeaways

  • AI design prioritizes user satisfaction via friction removal, which conflicts with developmental needs for struggle.
  • 'Struggle before assist' is a critical rule to preserve the learning process in education.
  • Synthetic validation can mimic friendship but lacks the reciprocity of human relationships.
  • Parents and educators should model behavior rather than relying solely on parental control software.

The Bottom Line

We spend our careers optimizing for latency and ease of use, but we need to remember that friction is a feature in human development. If we build tools that never say no, we risk raising a generation that doesn't know how to handle one.