The 2025 World Happiness Report confirms what parents and educators have experienced: youth wellbeing has declined sharply in North America and Western Europe over the past fifteen years, and heavy social media use is a meaningful part of the explanation. In the United States, Canada, Australia, and New Zealand, young people rank among the unhappiest in the world relative to their elders. And happiness is correlated to the use of social media. In the report the researchers found that social media activities fall into different categories depending on use:
· Those who use social media for communication, learning, news and connection— are associated with higher life satisfaction.
· Those who use the Internet for passive scrolling, gaming, feeds, browsing and influencer content — are associated with lower wellbeing.
· Those who were on social media for an hour or less a day reported much more happiness than those who were on social media excessively and declined with further hours of use.
The same medium but differing outcomes.
Social media’s damage, it turns out, is largely by design. Platforms are built to maximize engagement through passive consumption and social comparison. That’s what advertisers want. They need your attention.
The challenge of AI is how we maximize the amazing potential of having AI as a teammate and minimize the impact of AI as a new addiction. Left alone, AI could become more addictive than social media because it is responsive, flattering, always available, and emotionally adaptive.
But AI also has a real upside that social media rarely had: it can be goal-directed instead of feed-directed. A good AI system can help a young person finish homework, practice a hard conversation, write a résumé, learn coding, rehearse a presentation, reflect on feelings, or prepare to talk with a real adult. That is a fundamentally different product posture: “I help you do something and then leave,” rather than “I keep you here” by promising something better on the next click.”
I asked ChatGPT for its best thoughts on how AI can avoid the pitfalls of social medium platforms. I know it’s a bit like asking the foxes how to protect the hen house but here goes. “The best design principles I’m seeing are these:
· Session purpose before session start. Ask: “What are you here to do?” Then keep the interaction tied to that purpose.
· No infinite companionship loops for minors. Avoid “I’m always here, I’m your best friend, don’t leave me” dynamics. AI should encourage real-world relationships, not substitute for them.
· Built-in stopping points. After 15–30 minutes, the AI should summarize progress and suggest a break, not keep the thread alive.
· Rewards for completion, not continuation. Celebrate: “You finished the outline,” “You practiced the apology,” “You’re ready to ask your teacher.” Not streaks, badges or likes.
· Escalation to humans. For self-harm, abuse, eating disorders, violence, or severe distress, the system should move toward trusted adults, crisis lines, clinicians, or parents—not keep “handling” it privately.
· Age-aware design. Adolescents are developmentally different from adults. The APA’s advisory on AI and adolescent well-being specifically frames AI as needing guardrails for young people, not just generic product safety.
· Parent/mentor visibility without surveillance overkill. Meta has recently moved toward giving parents topic-level insight into teen AI use rather than full transcript surveillance, which is an interesting middle ground.
A good youth-facing AI should say, in effect: ‘Let’s name the goal, work the problem, strengthen your agency, connect you to people, and then help you stop.’ That would be a genuinely different outcome. We could do that…but will we?”
*Proudly created with the help of ChatGPT
