Pedagogical Agent for Children Emotions (PeACE)

Title: PeACE – Pedagogical Agent for Children’s Emotions

Project duration: 01.04.2026 – 31.03.2027

Research area: Responsible AI, early childhood education/development, Social-emotional learning

Uku is an AI conversational agent for social-emotional learning (SEL) in early childhood education, developed by PeACE Technologies, a spin-off from the University of Leipzig grounded in Prof. Dr. Tina Malti’s developmental psychology research (HumanKind lab). Children aged 3–6 interact with Uku in guided, one-on-one or small-group sessions moderated by educators, practicing skills such as recognizing emotions, empathy, and self-regulation through structured, playful dialogue. Uku is designed under a prevention instead of repair philosophy: building resilience early, before problems escalate, rather than intervening after difficulties arise. Crucially, Uku is built to complement, not replace, early childhood educators — offering consistent, scalable practice opportunities while leaving relational and pedagogical judgment to the trained professionals in the room.

Aims

Strengthen young children’s social-emotional skills early, building lasting resilience rather than addressing problems after they emerge. Provide educators with a scalable, evidence-based tool that extends their capacity without displacing their role. Validate Uku’s effectiveness and safety in real Kita settings through structured pilots, and grow toward broader adoption across German (and eventually international) early childhood education.

Problems

Social-emotional skills are among the strongest predictors of children’s long-term wellbeing, yet Kitas often lack time, staff, and standardized tools to foster them consistently. Existing AI-in-education tools tend to follow a replacement logic, raising concerns among educators and parents. The central question: how can an AI agent meaningfully support SEL in early childhood while strengthening — not undermining — the human relationships that are essential to children’s development?

Practical example

In a typical session, a child interacts one-on-one with Uku through guided, playful dialogue: identifying emotions in a scenario, naming feelings, and practicing a calming strategy such as breathing. An educator moderates the session, ready to step in for individualized scaffolding. This structure isolates Uku’s role as a consistent, standardized practice partner while preserving the educator’s central pedagogical function.

Technology

Uku is a conversational AI agent (voice- and dialogue-based) tailored for young children, designed and prompted around a structured social-emotional learning curriculum. It runs through a child-facing interface with an educator-moderated setup, and incorporates a voice/language stack currently being migrated to EU-based infrastructure for data protection alignment (GDPR/EU AI Act considerations).

Outlook

Building on this foundation, PeACE Technologies plans to launch structured pilots with Kitas in the Leipzig region to validate Uku’s effectiveness and safety in real-world settings, with Prof. Dr. Tina Malti advising on the research design. Sessions are planned to evolve from one-on-one interactions toward small educator-moderated groups of 4–5 children. Looking further ahead, findings from these pilots could inform a broader evidence base on how AI agents and educators complement one another in early childhood SEL — supporting expansion across German Kitas and, longer term, international markets, while contributing to the wider research conversation on responsible, human-centered AI in early education.

Publications

• Colasante, T. et al. (2026). Developing PeACE: A Pedagogical Agent for Children’s Emotions. In: Tammets, K., Sosnovsky, S., Ferreira Mello, R., Pishtari, G., Nazaretsky, T. (eds) Two Decades of TEL. From Lessons Learnt to Challenges Ahead. EC-TEL 2025. Lecture Notes in Computer Science, vol 16064. Springer, Cham. https://doi.org/10.1007/978-3-032-03873-9_40
• Raave, D. K., Colasante, T., Roa, E. R., Martinez, J. C. R., Li, H., Mukherjee, S., & Malti, T. (2026). An experimental study exploring human–AI complementarity in early social-emotional learning. Computers and Education Open, 10, 100331. https://doi.org/10.1016/j.caeo.2026.100331

Team

Lead

• Eric Roldan Roa

Team Members with Affiliation

• Doris Kristina Raave
• Anaïs Bourgon
• Johanna Eller,
• Ramos Martinez
• Juan Carlos Javier

Mentors

• Prof. Tyler Colasante
• Prof. Tina Malti

Partners

• Humboldt Wissenschaftszentrum für Kindesentwicklung
• SMILE/SEPT of Leipzig University