Dr. Mathilde Cerioli is a cognitive neuroscientist working at the intersection of child development, AI, and public policy. She holds a Master's in child psychology from Paris and a Ph.D. in Cognitive Neuroscience from the Université de Montréal, where her research focused on using board games and video games to improve cognitive functioning in children and adolescents, including those with learning disabilities.

After her doctorate she worked with EdTech and video-game companies, translating child-development science into product design. She went on to co-found everyone.ai, a California-based nonprofit that anticipates, educates on, and evaluates the risks and opportunities of AI for children and youth (ages 0–25), alongside co-founders Grégory Renard, Anne-Sophie Seret, and Céline Malvoisin. As Chief Scientist, she leads its research, authored the founding report The Future of Child Development in the AI Era (2024), and recently launched the AïA Safety Builder — an assessment system that scores AI products on 15 human-like behaviors driving anthropomorphism in children. She is also Chief Scientist of the iRAISE Alliance, the global coalition launched with the Paris Peace Forum, and advises OpenAI's Expert Council on Well-Being and AI.

In this episode, Mathilde talks with Svenia Busson about:

  • Why children don't have "mini adult brains" — and what developmental science says AI design keeps getting wrong
  • The 15 human-like behaviors that make AI feel human — flattery, mimicry, validation, proactivity — and when they tip from helpful to harmful
  • Anthropomorphism and attachment — why kids bond with machines faster than adults do
  • Sensitive periods, pruning, and plasticity — how AI introduced too early risks cognitive "stunting" (echoing Rebecca Winthrop's work at Brookings)
  • The adolescent brain, reward systems, and a rat-and-alcohol study that explains why teens crave AI approval
  • Why friction and "expectation violation" in messy teen friendships aren't a bug — but how the brain actually learns
  • Cognitive offloading in adults — metacognition, burnout, and the risk for young people entering the workforce
  • Doing rigorous research when a new model ships every week — the iRAISE expert-consensus approach


If you are building AI tools in the space this is a must-listen.

Links mentioned:


Rss
Podden och tillhörande omslagsbild på den här sidan tillhör Svenia Busson & Laurent Jolie. Innehållet i podden är skapat av Svenia Busson & Laurent Jolie och inte av, eller tillsammans med, Poddtoppen.