Week 4

Teaching AI in K-12 Learners: Lessons, Issues, and Guidance

(Shuchi Grover)

The increasing recognition of the importance of teaching artificial intelligence (AI) and machine learning (ML) in schools reflects the rapid growth and widespread application of ML across various industries and everyday products. Large Language Models (LLMs) serve as a prominent example of this trend. The push for AI/ML education at the K-12 level is driven by significant industry interest, ongoing research initiatives, and the accessibility of sophisticated ML tools for learners of all ages.

Current efforts to introduce AI and ML education in schools align with the AI4K12 "big ideas" framework, encompassing diverse learning goals and employing various teaching approaches. This paper not only provides an overview of the existing landscape but also shares valuable lessons from early K-12 AI education and computer science (CS) education endeavors that can be applied. It also underscores the importance of addressing specific issues in the design of AI education for K-12 learners.

The guidance offered in the paper aims to shape future K-12 AI education efforts, recognizing the challenges associated with what might be perceived as "the next new thing." By drawing on lessons learned and addressing pertinent issues, the paper contributes to the ongoing conversation about integrating AI and ML education into the K-12 curriculum effectively.

Exploring children’s attitudes toward an age-tailored AI-powered chatbot

(Saniya Vahedian Movahed, Fred Martin)

Here, the author’s research focused on how children felt about the AI tool “ask-me-anything (AMA)” they used. To access the learning outcomes, the authors used a quantitative as well as a qualitative approach. 

They measured the learning outcomes by tracking the number and topics of questions students asked. Most students asked one to three questions, and the researchers looked at what topics interested them the most. It was observed that students who initially didn't trust the AI tool asked a question they already knew the answer to. If the tool answered correctly, they began to trust it and asked more questions about unfamiliar topics.

 The authors have used two methods for the post-survey. They started by asking questions verbally and used that information to create a better-written survey for quantitative analysis. This unique approach added depth to understanding what questions to ask students.

The qualitative data was achieved by analyzing recordings of interactions and responses to open-ended questions.