AI-Enabled OER Commons Support for Students with Learning and Cognitive Print Disabilities

Link to this OER Commons Group

Goals

Objective: Mitigate/eliminate issues associated with LLMs

 

 

Link to this OER Commons Group:

https://oercommons.org/groups/ai-enabled-oer-commons-support-for-students-with-l/14795/

This group investigates the advantages of employing IDEAL Group’s Textbook-Based Model (TBM), which adapts open textbooks from OER Commons (and other OER resources) to produce tailored content for students with learning and cognitive print disabilities. This method offers a sophisticated alternative to using Large Language Models (LLMs) to create educational materials.

Goals:

1. Collaborative Development of TBM Queries: Members will collaborate to discuss, develop, and disseminate TBLM queries that generate content specifically tailored to strengthen individual cognitive abilities as outlined in the Cattell-Horn-Carroll Framework (CHC)
   http://www.iapsych.com/chcdefs.pdf;
2. Create Universal Design for Learning (UDL) Compliant Content: The effort will center on crafting content that adheres to the Universal Design for Learning framework guidelines, ensuring that educational materials are beneficial and accessible to all students.
   https://udlguidelines.cast.org/; and,
3. Utilization of Open Educational Resources (OER): The initiative will involve using Open Educational Resources (OER) as foundational material for constructing and refining TBMs.
   https://oercommons.org/

Objective: Mitigate/eliminate issues associated with LLMs.

1. Inference Limitations: Vetted textbooks present information in a structured, pedagogically sound manner, which simplifies the process of inferring main points and providing coherent summaries.
2. Knowledge Cut-off Dates: Textbooks, especially those covering foundational knowledge, tend to remain relevant longer and are updated less frequently, enhancing the model’s long-term accuracy.
3. Training Data Limitations: Unlike broader models, TBLMs built on current textbooks and educational materials sidestep outdated training data challenges.
4. Bias and Fairness: Textbooks employed in classrooms generally undergo thorough reviews to ensure accuracy and fairness, thus reducing potential biases.
5. Misinformation and Disinformation: The rigorous review process for approved textbooks diminishes the likelihood of misinformation and disinformation.
6. Contextual Misinterpretation: While not foolproof, the structured nature of educational texts helps lessen contextual misinterpretations.
7. Adaptability to Curriculum Changes: Although TBLMs cannot dynamically update to reflect the very latest curricular changes, they generally conform to current educational standards.
8. User Interaction Limitations: Despite inheriting some interaction limitations of LLMs, precise queries and directives can enhance interaction effectiveness.
9. Hallucination of Facts: Using vetted educational materials reduces the risk of the model generating inaccurate information.
10. Incoherent or Irrelevant Responses: The clarity and organization of textbook content generally improve the relevance and coherence of responses.
11. Intellectual Property and Copyright Issues: Employing resources from OER Commons typically resolves many intellectual property concerns, assuming adherence to the platform’s guidelines.