AI and Child Development Researcher at Everyone.ai. Cognitive Sciences Master’s student. Switzerland.

Olga Muss Laurenty is an AI and child development researcher whose work integrates engineering, cognitive science, and a multicultural perspective gained from living in five different countries. Olga is based in Switzerland and also works with California, U.S. and is a co-author of the 2024 report “The Future of Child Development in the AI Era.”

Description of her work

• With everyone.ai, Olga works on projects examining how large language model (LLM) design can mitigate negative impacts of AI on child development. This involves prioritizing emotional reliance, addressing sycophancy, and other manipulative techniques.

  * Sycophancy is a pattern where an AI model prioritizes aligning its responses with the user’s beliefs or opinions over providing accurate, truthful information.

• Olga collaborates with a-i.swiss, HEP Vaud (a teacher preparation school in Switzerland), and ETH Zurich to design and develop solutions to support the safe and adapted use of AI in education. These frameworks extend beyond traditional chatbots by enabling monitoring of understanding, well-being, and task focus.
• At ETH Zurich, Olga works with Luca Leisten in the Social Brain Sciences Lab to develop a curriculum for German STEM classes for 14–15-year-olds, emphasizing the development of social robots with safe LLMs and advancing AI literacy.

Key findings

• There remains significant uncertainty around many aspects of child development.

• Child development and social dynamics are influenced by many interrelated factors that don’t work in a simple straight-line cause-and-effect way. Instead, they involve ongoing cycles of positive and negative feedback, which means some patterns can reinforce themselves while others stabilize. Studying these cycles helps us understand developmental and social dynamics.

• When designing programs or technologies for children, it’s essential to consider how their thinking develops and their biological constraints, such as sensitive periods, developmental stages, and emotional needs.

• For solutions to genuinely help children, they need to have a clear purpose and direction, so they shape development in positive ways.
• The need for rigorous, collaborative, reproducible research is urgent. Artificial intelligence offers promising tools to speed up this work, and the risks of AI in this field seem manageable.