ml-machine.org: Infrastructuring a Research Product to Disseminate AI Literacy in Education

Title of the Paper

ml-machine.org: Infrastructuring a Research Product to Disseminate AI Literacy in Education

Paper Information

• Subject Area: Human-Computer Interaction, Educational Technology, AI Literacy
• Keywords: AI literacy, research product, participatory infrastructuring, educational tool design, human-computer interaction, K-12 education, machine learning teaching, technology dissemination, education scalability, school technology integration

Research Background and Problem

• Problems or Challenges:

  1. Current AI literacy educational tools face limitations in impact and adoption, with limited research exploring how these tools can be integrated into real-world educational practices on a large scale.
  2. Many existing teaching resources have high technical barriers, are costly, and lack discussions on how AI works, focusing instead on demonstrating its capabilities.
  3. There is a general lack of long-term research and strategies for integrating AI tools into K-12 educational curricula.

• Significance:

  • With the rapid development and widespread application of artificial intelligence and machine learning, fostering AI literacy among students is increasingly important for school education and societal transformation. The lack of educational tools that can be effectively integrated into classrooms and adopted by teachers limits the promotion of AI education.
  • AI education needs to focus not only on technical content but also on guiding students to reflect on AI's real-world societal impacts.

• Research Motivation and Related Work:

  • Current HCI research on AI literacy education primarily focuses on developing prototypes, with limited studies on practical applications at scale.
  • This paper aims to explore how to embed educational tools into teaching practices and educational infrastructures while ensuring their long-term impact and adoption by developing ml-machine.org as a research product.

Solution

• Proposed Approach:

  • Develop a web-based and Micro:bit-based educational tool called ml-machine.org to teach K-12 students about machine learning concepts and techniques.
  • Employ a Research Product Approach and Participatory Infrastructuring, collaborating with multiple strategic partners in education and technology to ensure the tool's dissemination and integration into formal education systems.

• Innovations:

  1. Research Product Framework: Transform temporary teaching prototypes into mature products suitable for classroom adoption and scalable use.
  2. Embedding in Educational Infrastructure: Build partnerships to integrate the tool with existing educational resources (e.g., Danish Broadcasting Corporation course materials and Micro:bit hardware), reducing barriers to educational use.
  3. Openness and Scalability: Emphasize low-cost implementation (web-based and Micro:bit architecture) in the tool's design, supporting teachers and partner organizations in expanding its use.

• Implementation Steps:

  1. Tool Design: Develop software that supports teaching through embodied exploration and interactive activities, enabling students to collect data with Micro:bit, create simple ML models, and conduct experiments.
  2. Technical Enhancements: Simplify hardware-software connection processes by introducing Web USB and Web Bluetooth technologies, lowering technical barriers.
  3. Collaborative Dissemination:
     • Collaborate with the Danish Broadcasting Corporation (DBC) to design curriculum units.
     • Partner with the Micro:bit Educational Foundation (MEF) for open-source development, software maintenance, and international promotion.
  4. Embedded Scenario Exploration: Adjust and promote the tool based on real feedback from teachers, educators, and students.

Research Outcomes

• Specific Outcomes:

  1. User Scale: ml-machine.org and its customized versions have attracted over 5,000 unique users (as of February 2024).
  2. Teacher Adoption and Feedback: Multiple teachers and educational consultants have adopted and shared the tool. Some have adapted its content into lesson plans, while others have designed new activities for other teachers and students.
  3. Partnerships: Achieved deep collaboration with leading educational institutions and technology organizations (e.g., DBC and MEF) and secured new funding for continued development.

• Comparison with Existing Solutions:

  1. Greater focus on scalable adaptation and long-term impact of tools rather than short-term teaching prototypes.
  2. Provides a low-barrier, open-source, and reliable hardware and software framework.
  3. Bridges the gap between academic research and practical applications by leveraging educational networks and public media for impact dissemination.

• Experimental and Evaluation Results:

  • The tool's flexible design received high praise from teachers for its ease of use, quick activity model setup, and emphasis on student experience and reflection.
  • Teachers also expressed a desire for more advanced features to support diverse teaching needs.

• Limitations and Future Directions:

  1. Data Collection Limitations: The adoption and usage of the tool cannot be tracked in fine detail, partly due to privacy regulations, making it difficult to fully reveal usage scenarios.
  2. Teacher Support Needs: Teachers with different disciplinary backgrounds vary in their understanding and application of the tool, necessitating more systematic support solutions such as teacher training and clearer teaching guidelines.
  3. Long-term Research Plans: In-depth research on the long-term adoption process by teachers has not yet been fully initiated, though educational organizations have begun proactive efforts in this area.

Conclusion and Insights

• This study combines the Research Product Approach and Participatory Infrastructuring to address the challenges of adoptability and sustainability of AI literacy educational tools in K-12 contexts.
• It introduces "infrastructure" as a research product attribute, emphasizing how technological tools can integrate into educational ecosystems and existing practices.
• Future work will focus on the global expansion of teaching tools and the long-term tracking of the teacher adoption process, providing data and practical support for further optimizing AI education.