Governments Advance AI in Schools Legislation
Multiple governments are enacting new rules for AI in education. In the U.S., a bill to improve AI safety education in schools is advancing to the Virginia Senate, while an Arizona House committee backed a bill requiring K-12 AI curriculum. The UAE has banned generative AI tools for students under 13, citing safety concerns.
- The Virginia bill, HB 171, was introduced by Delegate Alex Askew and specifically aims to update existing internet safety policies to include instruction on AI-generated content, online scams, and misinformation. This follows a 2024 executive order that laid out guidelines for the responsible integration of AI in education. - Reinforcement learning techniques, such as Q-Learning, can be used to create adaptive learning systems that personalize educational content in real-time based on a student's performance and interactions. This approach models content delivery as a sequence of decisions to maximize learning engagement and comprehension. - Knowledge Tracing is a student modeling technique used in intelligent tutoring systems to infer a learner's knowledge state over time. Models like Bayesian Knowledge Tracing (BKT) and Deep Knowledge Tracing (DKT) use a student's interaction data to predict their future performance on learning tasks. - Speech recognition for young children presents unique challenges due to their higher-pitched voices, shorter vocal tracts, and less predictable speech patterns, such as over-pronunciation and creative grammar. Effective systems require specialized models trained on large datasets of children's speech to accurately assess phonemes, fluency, and pronunciation. - Multi-armed bandit (MAB) algorithms offer an alternative to traditional A/B testing for recommending educational content. MABs can more efficiently determine the most effective content by dynamically allocating more "traffic" to options that show higher engagement, balancing exploration of new content with exploitation of known effective content. - While the UAE has restricted generative AI for younger students, it is also implementing a comprehensive AI curriculum starting in kindergarten. The curriculum is designed to build AI literacy by teaching concepts like algorithms, bias, ethics, and even prompt engineering in later grades. - User experience (UX) design for young children requires special considerations, such as large, easy-to-tap buttons (at least 48x48 dp), minimal text, and simple navigation with a flat hierarchy. Using bright colors, consistent visual cues, and sound feedback can help maintain engagement and provide necessary feedback for non-readers. - Broader legislation like the Kids Online Safety Act (KOSA) and the California Age-Appropriate Design Code are pushing platforms to assess and mitigate risks driven by AI, which will impact how educational AI tools are designed and deployed.
