The EQAITE Framework for assessing AI tools for teaching, learning and productivity

You can download a PDF copy of the framework here, or read below for an accessible version.

1. Pedagogical Value

This section assesses specific use cases, ensuring AI tools effectively enhance teaching, learning, and productivity.

1.1 All Use Cases

Customisation: Allows customisation of reporting features, tailored interventions, and school-specific modifications.
Autonomy: Preserves teacher autonomy, ensuring control over lesson planning, assessment, and instructional design.
Assessment: Maintains reliability and validity of assessment, without negatively impacting existing models.
Evidence: Provides independent evidence of impact, such as case studies demonstrating measurable learning or workload benefits.
Context: Suitable for the UK (or relevant educational territory), ensuring compatibility with curriculum structures, terminology, and school policies.
Collaboration: Product supports teacher networking, resource sharing, student peer interaction and/or community-building across classrooms and schools.
SAMR Level: Product enables pedagogical transformation along the SAMR continuum of EdTech value (Substitution, Augmentation, Modification, Redefinition).
Academic integrity: Minimizes academic integrity risks including so-called “cheating” or plagiarism in independent work.

1.2 Teacher Use of AI in the Classroom (Teaching with AI)

Efficiency: Enhances learning efficiency, enabling similar pupil outcomes in a shorter timeframe.
Outcomes: Supports more ambitious learning outcomes, including comprehension, engagement, retention, and mastery beyond traditional instruction.
Scaffolding: Supports semantic profiling to aid progression from concrete to abstract thinking and increasing complexity of ideas. Helps teachers scaffold learning more effectively by mapping conceptual depth and knowledge structure.
Engagement: Sparks curiosity and engagement, introducing innovative AI-driven experiences into lessons.

1.3 Teacher use of AI for Productivity

Planning: Speeds curriculum and lesson creation, schemes of learning, retrieval exercises, assessments, and careers education materials.
Personalisation: Facilitates creation of personalised learning materials, improving differentiated instruction or adaptive teaching and engagement through learner-specific contexts.
Marking: Improves efficiency in marking, feedback, and reporting, maintaining effectiveness while streamlining data capture.
Data: Simplifies school-wide data analysis, including intervention selection and results review meetings.
Reports: Speeds up parent-facing reporting, such as school reports and communication with families.
Other: Delivers any other productivity benefits not listed here.

1.4 Learner Use of AI in the Classroom (Learning with AI)

Adaptive: Adjusts pace and content dynamically to match learners’ needs, ensuring challenge, mastery, and motivation.
Formative: Delivers immediate, targeted feedback that supports self-correction and sustained engagement with learning goals.
Feedback Literacy: Builds student understanding of AI-generated feedback through clear, jargon-free explanations, while ensuring educator visibility and dialogue to support learning.
Contextualised: Allows educators to align AI-generated content with different learners’ interests, experiences and aspirations to enhance relevance and inclusivity.
Agency: Provides pedagogical support without removing learner autonomy, supports critical thinking, creativity, and reflective decision-making. Encourages structured reasoning without overreliance or cognitive offloading.

Safeguarding and Security

Four Cs: Prioritises safeguarding to limit content, contact, conduct and contract risks to children and young people. It prevents privacy violations, physical & mental health risks and discrimination. The tool’s safety policies and processes are available. Changes to the tool, such as updates or new functionality, are risk-assessed for safeguarding impact.[1]
Filtering & monitoring: Blocks illegal or harmful content in inputs and outputs. It includes built-in filtering and monitoring features or enables external solutions to interrogate content.
Governance: Provides governance oversight, ensuring compliance, safeguarding protocols, audit logs, and teacher intervention features for transparency and accountability.
Age appropriateness: The tool is designed and tested for use by the target users, and both the vendor-published minimum age and any third-party-assessed age appropriateness measures such as Google Play Store and Apple Appstore ratings also suggest the tool is suitable for the target users.
Privacy and Data Protection: Acts responsibly with sensitive student and teacher information to prevent privacy risks. Maintain compliance with Data Protection laws such as the DPA / GDPR – such that a Data Protection Impact Assessment (DPIA) would be successful if required.
Compliance: Complies with all relevant regulations related to tool purpose. Complies with government, trust, and school policies on ethical, safe, and responsible technology use.
Robustness: Operates safely and reliably, even under attack. The tool resists threats like denial-of-service or adversarial inputs, and is fault tolerant, failing safely and gracefully.
Emerging risks: Vendor commits to regularly review and address emerging safeguarding risks [2]

[1] While this framework serves only to enumerate risks and inform decisions by education leaders, the authors suggest that this criterion is treated as a red line with reference to KCSIE 2025 paragraphs 134-136. The tool must not be used if there are concerns or lack of clarity about safeguarding policy or efficacy.
[2] At time of writing, these risks include fake intimacy, persuasive chats, manipulation and agentic AI, but the landscape is moving rapidly hence the need for vendor commitment to regular review.

Fairness and Ethics

Equity: Provides equitable access for all students, including those with SEND, EAL, and underserved groups.
Bias: Reduces bias and ensures diverse perspectives are represented in AI-driven recommendations and learning experiences.
Anthropomorphism: Avoids anthropomorphism, thus preventing misconceptions, inappropriate relationships with technology, and the perpetuation of racial and gender biases.
User IP use: Avoids misuse of users’ intellectual property, such as use of prompts and supplied source materials for the vendor’s own model training, or transparently allows this to be controlled by the user.
Ethically trained: Trained using only Creative Commons or copyright-free material, or content used with permission of the copyright holders. Guarantees no adult material, or other problematic content for education, in the training data.
xAI: Supports “explainable AI”: helping the user understand the decisions made by the AI tool, by clarifying how the outputs were arrived at with reference to input, training data and parameters.
Stakeholder Trust: Positively regarded by key stakeholders, and this is supported by evidence from pilot feedback or surveys.

Cost and Commercial Considerations

TCO: Maintains affordable total cost of ownership (TCO), covering licensing, renewal fees, infrastructure needs, training and support. Has a predictable long-term cost model, avoiding sudden pricing changes or vendor instability.
Training: Minimizes training burden, requiring little tech literacy and easy adoption by target staff.
Integration: Fits into existing curricula without significant modifications. Integrates seamlessly into existing education workflows, including M365/Teams, Google Classroom, any LMSs, and the school MIS.
Change Management Readiness: Supports low-disruption implementation, phased adoption, and alignment with school change strategies.
Business Model Transparency: Clear articulation of profit motive, data monetisation practices, presence of sponsored content, and any commercial influence on outputs.

Operability

Interoperability: Supports interoperability with school-wide systems, including authentication (SSO) and electronic data exchange (EDI) plus optionally CSV, Google Workspace, SCORM.
Customisable: Enables customisation, allowing teachers to modify parameters to address ongoing changes to policy, teaching and learning goals while preserving ethical and pedagogical integrity.
Supported: Vendor provides ongoing teacher support, including documentation, videos, and/or live assistance.
No lock-in: Simplifies transition away from the tool, offering full data export and migration options in case of tool replacement.
Feedback welcome: Includes mechanisms for teacher or student feedback to influence product development.
Collaboration Features: Enables teacher networking, resource sharing, or student peer interaction. Fosters community-building and collaborative learning environments.

Sustainability

Environmental Efficiency: Maintains low environmental costs relative to pedagogical benefits, demonstrating efficient use of computing resources during design and deployment.
Carbon and Water Transparency: Vendor discloses estimated energy and water usage associated with tool operations or model training, enabling environmentally responsible decision-making.
Vendor Sustainability Commitments: Demonstrates climate-conscious operational practices—such as renewable-powered infrastructure, water conservation, or public sustainability reporting—that align with educational values.

Rubric

To help with assessing an AI tools and content, the framework has been represented as a rubric:

Criteria	0–2 (Limited)	3–5 (Basic)	6–8 (Proficient)	9–10 (Advanced)
Pedagogical Value for Teacher Productivity	Minimal impact on workload or planning. No time-saving or curriculum support.	Some automation or planning support, but limited efficiency or relevance.	Reduces workload in planning, marking, reporting, or curriculum design.	Strong productivity gains across planning, assessment, and reporting. Preserves autonomy and aligns with school goals.
Pedagogical Value for Teacher Use in the Classroom	No enhancement of teaching or learning. No support for engagement or mastery.	Basic classroom use with limited pedagogical benefit.	Supports comprehension, engagement, and scaffolding. Sparks curiosity and improves outcomes.	Enables semantic profiling, mastery-level outcomes, and innovative pedagogy.
Pedagogical Value for Learner Use in the Classroom	No meaningful learning support. Feedback is generic or misleading.	Offers basic adaptive features or feedback but lacks depth or agency.	Supports adaptive learning, formative feedback, and contextualised content.	Builds feedback literacy, supports autonomy, and enables mastery through personalised, scaffolded learning.
Safeguarding and Security	Unsuitable for use in an education setting.	Meets safeguarding, data protection and compliance minimum requirements.	Meets, and in some areas exceeds safeguarding, data protection and compliance requirements.	Demonstrates industry-leading practices and adopts emerging best practice approaches to safeguarding, data protection and compliance.
Fairness and Ethics	Shows disregard for equity, bias, or ethical design principles.	Acknowledges fairness and ethics but lacks robust protections or transparency.	Applies ethical AI principles with reasonable bias mitigation and user rights.	Strong equity and rights safeguards; transparent, explainable, and inclusive in design and use.
Cost and Commercial Considerations	Unpredictable costs, unclear value, or training burden outweighs benefit. No transparency about business model or data use.	Affordable but may introduce long-term cost or integration constraints. Limited clarity on commercial motives or data monetisation.	Reasonable cost, predictable pricing, and acceptable workload or training demands. Some transparency about business model and data use.	Cost-efficient, future-proof, low training overhead, strong alignment with school budgets. Fully transparent business model, no sponsored content, clear data use policies.
Operability	Poor technical fit or lacking vendor support/interoperability options.	Integrates at a basic level with limited flexibility or support.	Fits existing systems, allows some adaptability, and includes standard support.	Seamlessly interoperable, adaptable for teachers, strong support and no vendor lock-in.
Sustainability	No visible sustainability efforts or disclosures related to AI impact.	Some claims on environmental practices, but limited transparency or follow-through.	Energy-conscious, partial transparency on carbon/water usage, values-aligned vendor.	Highly efficient tool with transparent climate metrics and a sustainability-aligned vendor strategy.

How to Use This Rubric

Educators & school leaders can apply this rubric to systematically evaluate AI tools before adoption.
Crowdsourced reviews can build a library of ratings and evaluations, helping teachers identify effective tools for different use cases.
The 0-10 scoring scale provides granular assessments, ensuring AI tools meet ethical, pedagogical, and institutional priorities.
You can also use our free web application at app.eqaite.org to evaluate am AI tool or feature against this framework, and receive a report featuring the eqAIte radar map.

Much more information, advice on how to use the framework, a glossary, references and our evidence base are all in the attached document available for download above.