"I Can Be Anything!" Bridging Today and the Future through Generative AI-driven Self-represented Career Imagination for Children

Paper Title

"I Can Be Anything!" Bridging Today and the Future through Generative AI-driven Self-represented Career Imagination for Children

Publication Info

• Topic area: Generative AI applications in child development and career imagination.
• Keywords: Generative AI, career imagination, self-continuity, children, multimodal large language models, text-to-image models, identity development, storytelling, future self, child-centered design.

Background and Problem

• Problem / challenge: Existing career imagination interventions for children often fail to connect their present identity with future aspirations, treating careers as static labels rather than dynamic, self-relevant narratives. Current tools are also largely designed for adolescents or adults, assuming prior knowledge and abstract reasoning abilities that younger children lack.
• Significance: Cultivating a sense of self-continuity in middle childhood is critical for identity formation, motivation, and long-term goal setting. Addressing this gap can help children envision meaningful futures and align their present learning with future aspirations.
• Motivation and related work: Prior interventions include arts-based activities, role-play, and gamified career exploration, but these often lack personalization and fail to engage younger children effectively. Generative AI offers new opportunities for creating personalized, multimodal experiences, yet its application in child-centered career imagination remains underexplored.

Solution

• Proposed approach: FutureMe, a generative AI–driven system combining multimodal large language models (MLLMs) and text-to-image (T2I) models to help children aged 7–11 imagine and narrate their future selves in career contexts.
• Novelty:
  1. Introduction of a system that integrates real-time photo capture, contextual image editing, and narrative co-construction to operationalize self-continuity.
  2. Demonstration of how generative AI enables children to disclose hidden aspirations, connect past and present experiences to future careers, and reinterpret externally imposed career norms.
  3. Design implications for creating child-centered AI tools that foster agency, personalization, and reflective engagement.
• Procedure and key techniques:
  1. Career Exploration: Children browse a "career universe" of 50 diverse professions visualized through T2I models.
  2. Cartoonize: A child’s photo is transformed into a cartoon-style image to reduce self-consciousness and enhance privacy.
  3. Careerize: The system generates three personalized career images based on the child’s chosen profession, allowing them to select their preferred visualization.
  4. Describe: MLLMs generate child-friendly explanations of the chosen career.
  5. Storize: Children define actions they would perform in their future role, which are transformed into three-panel story strips.
  6. Postcard Output: A tangible postcard summarizing the child’s present self, future self, story strip, and a reflective letter to their future self.

Results

• Concrete findings:
  • 10 out of 17 children who initially reported no specific career aspirations developed new, concrete dreams after using FutureMe.
  • 53% of participants demonstrated vividness in imagining future roles, 41% showed identity similarity, and 35% expressed emotional connectedness to their future selves.
• Advantage over baselines:
  • Enabled children to move beyond static job labels by creating dynamic, self-represented narratives.
  • Fostered self-continuity and agency, helping children reinterpret externally imposed career expectations into personally meaningful futures.
• Experiments / evaluation:
  • Conducted a workshop with 17 children aged 7–11, involving one-on-one sessions lasting 45–60 minutes.
  • Data sources included visual narratives, handwritten letters, and qualitative interviews.
  • Mixed-method analysis revealed shifts in children’s imagination across three dimensions: vividness, identity similarity, and emotional connectedness.
• Limitations and future work:
  • Technical limitations include occasional misalignment between generated outputs and children’s intent, and latency in the generation process.
  • The study's small sample size and single-session design limit generalizability; future work should include longitudinal studies and larger, diverse samples.
  • Ethical concerns around embedding children’s faces in AI-generated content require further exploration.

Summary

FutureMe is a generative AI system designed to help children aged 7–11 imagine and narrate their future selves in diverse career contexts. By integrating multimodal large language models and text-to-image generation, the system enables children to explore careers, visualize themselves in professional roles, and create personalized narratives. A study with 17 children demonstrated how FutureMe fostered self-continuity, revealed hidden aspirations, and empowered children to reinterpret externally imposed career norms. While the system shows promise in supporting identity development and career imagination, future work should address technical, methodological, and ethical limitations to enhance its scalability and impact.