AI Use Overview

How AI Is Defined in This Playbook

Using the Alan Turing Institute’s definition, AI refers to the design and study of machines that can perform tasks previously requiring human intelligence. 

• Machine Learning is a subset that uses algorithms to identify patterns in training data and apply those patterns to make predictions or informed estimates about new data (IBM).
• Generative AI (GenAI) is a subset that uses machine learning to generate new content—conversations, text, images, or recommendations—that resembles training data (CoSN GenAI Readiness Checklist).
• Large Language Models (LLMs) is a type of GenAI that is trained on massive amounts of data using advanced machine learning to learn patterns in human language and generate responses based on written prompts (Stanford University IT).
• Natural Language Processing (NLP) is a type of AI that focuses on how computers understand, interpret, and respond to human language, helping systems make sense of text or speech and generate responses that are clear and meaningful (ISO).
• Computer Vision is a type of AI that allows computers to process and understand visual data using techniques like machine learning to recognize patterns in visuals, including objects, scenes, and movement (Digital Promise Glossary of AI Terms for Educators).

Because AI processes data according to the data and/or prompts it receives, humans play a critical role in determining what to ask of AI and evaluating the accuracy of its patterns and outputs.

Considerations for AI Use & Privacy

• Understand district policies: Review your Acceptable Use Policy (AUP) or Responsible Use Policy (RUP) and Code of Conduct for guidance on AI use and compliance tracking. If your district has not adopted a policy, check your state board of education or state school board association’s website for guidance.
• Evaluate privacy terms: Find out what data is collected, how it is stored and for how long, how it may be used (including for model training), whether it is shared with third parties, and what control you have over your data. Leverage field standards by confirming whether the product has earned certifications through Digital Promise’s Responsibly Designed AI product certification, Common Sense’s Privacy Rating, or 1EdTech’s Data Privacy Seal. 
• Know your data governance: Understand how data governance policies affect AI use for your team. 
• Evaluate data security: Learn the product’s breach response plans and compliance with relevant laws (FERPA, COPPA, etc.). Leverage iKeepSafe’s COPPA Safe Harbor and FERPA Certifications. 
• Understand bias mitigation strategies: Determine how the product is intentionally identifying and addressing instances of bias in the model, and how those instances can be reported to edtech providers. 
• Provide detailed prompts: When using AI, include relevant context and explicitly request that the AI name any omitted or missing data.
• Identify AI: Learn how the tool indicates when a user is interacting with AI or accessing AI-generated content. 
• Establish review processes: Collaborate with your team to review AI-generated content for accuracy and bias. Ensure humans make final decisions for each student.
• Communicate with families: Use district or state policy to guide how you communicate AI use in student success work.
• Maintain human override: Contact relevant providers to learn how humans can override AI-driven recommendations.

For more on these considerations, see CoSN’s AI Readiness Framework and Digital Promise’s Responsibly Designed AI Certification and AI Literacy guidance.

The Four Pillars of Student Success Systems

The GRAD Partnership advocates for student success work to be built on four essential pillars:

1. Strong, Supportive Relationships Success requires cultures of belonging and proactive relationship-building across school staff, students, and families. Teams should reflect on which data support these relationships, co-create action steps with stakeholders, and commit to continuous improvement).
2. Real-Time, Actionable, Holistic Data Teams need access to timely, research-based indicators—including attendance, behavior, course performance, and measures of student agency, belonging, and connectedness—to identify students who may be falling off-track and intervene early.
3. Strategic Improvement Actions Regular review of data enables teams to develop meaningful, student-centered actions. This includes assigning appropriate interventions, tracking implementation, and building institutional knowledge about what works for which students.
4. Student-Centered Mindsets Shared beliefs about student success must center respect for the whole child. Through dialogue, reflection, and shared experiences, teams can identify and address systemic barriers to graduation 

(GRAD Partnership, 2024)