AI, Ethics & Responsible Use
How we use AI in GCSE Python Lab, the guardrails we've built, and our commitment to responsible educational technology
Our Approach to AI
GCSE Python Lab uses artificial intelligence to reduce teacher workload and provide instant feedback to students. But we're not naive about the challenges and responsibilities that come with educational AI.
As a teacher-built platform, we've designed our AI features with a clear principle: AI should support learning, not replace teaching.
This page explains exactly how we use AI, what safeguards we've built, and how we comply with UK regulations on AI in education.
How We Use AI
1. Student Support (Hints & Guidance)
When students get stuck, they can request hints at four levels: Low, Medium, High, or Step-by-step.
What the AI does: Provides guidance using the Socratic method—asking questions, pointing to relevant concepts, suggesting approaches.
What the AI does NOT do: Give complete answers, write full code solutions, or do the work for the student.
2. Automated Marking
For certain question types—especially coding exercises, trace tables, and written answers—AI helps grade submissions using exam-style rubrics.
What the AI does: Applies mark schemes, awards marks for correct answers or approaches, provides structured feedback on strengths and areas to improve.
What the AI does NOT do: Make final decisions. Teachers can review, adjust, and override all AI-assigned marks. For high-stakes assessments, teachers should always verify marks before releasing to students.
3. Question Generation (Teacher Tools)
Teachers can use AI to help generate new questions based on topics and difficulty levels.
What the AI does: Suggests question ideas, creates starter code, writes mark schemes.
What teachers do: Review, edit, and approve all AI-generated questions before assigning to students. Teachers maintain full editorial control.
AI in Content & Platform Development
📝Question Bank Creation
The 500+ question bank was created using AI assistance to speed up content development. Questions were generated based on my own teaching materials from 10+ years in the classroom, with human oversight throughout.
Important: I did not personally train AI on OCR past papers, mark schemes, or specification text. I did not copy and paste from copyrighted exam board materials. Questions are original content aligned to the J277 specification—the same approach as writing my own teaching resources, just faster with AI help.
However: As with all AI generation, I cannot guarantee that the AI models I used (such as GPT-4) were not trained on OCR or other exam board materials as part of their large-scale training datasets. Due to the size of the question bank, it is not practical to cross-check every question against all existing exam board materials.
All questions are OCR-style (following the J277 specification) but are not OCR questions. They are original works created with AI assistance and informed by my teaching experience. If you have copyright concerns about any specific question, please contact me and I will review and remove it if necessary.
💻Website Development
As a Computer Science teacher with a degree in Mathematics and Computer Science, I built this platform using Node.js/TypeScript with integrations into Google Firestore and OpenAI. The development process used modern software development tools and AI-assisted code generation (GitHub Copilot, Claude).
AI assistants wrote significant portions of the code—particularly UI components, boilerplate, and routine implementations. The system architecture, feature design, pedagogical decisions, and technical oversight are mine. AI acted as a coding partner, significantly speeding up development while I directed what to build and reviewed what it produced.
As a full-time teacher with limited spare time, AI-assisted development has been the most practical way to build a production-quality platform that would otherwise require months of full-time work.
Guardrails & Safety Mechanisms
We've built multiple layers of protection to ensure AI is used responsibly:
🛡️ System-Level Prompt Guardrails
Every AI interaction includes a system prompt that explicitly forbids:
- Revealing full working solutions
- Writing complete code for the student
- Doing the work instead of guiding learning
These guardrails are enforced at the system level and cannot be bypassed by student prompts like "just give me the answer."
📊 Rate Limiting
To prevent over-reliance on AI:
- Per-user daily limits: Students can only request a certain number of AI hints per day
- Per-submission limits: Limited feedback requests per assignment submission
- Test mode restrictions: AI help is completely disabled during formal assessments
👨🏫 Teacher Override & Control
Teachers maintain final authority:
- Review and adjust any AI-assigned marks
- Add manual feedback alongside or instead of AI feedback
- See which questions were AI-graded vs auto-graded
- Disable AI features per assignment (e.g., for formal tests)
- Control when marks are released to students
🚫 Anti-Cheating Detection
Our marking AI detects blatant attempts to game the system:
- Answers like "give me 10/10" or "I deserve full marks" receive 0 marks
- Normal academic language is NOT penalized
- Suspicious attempts are logged for teacher review
🔒 Data Protection & Privacy
When using AI features (powered by OpenAI):
- We send minimal data—question text and student's submission only
- No personally identifiable information (names, emails) is included in AI requests
- OpenAI does not use our data to train their models (per API agreement)
- All data transmission is encrypted (HTTPS/TLS)
- See our Privacy Policy for full details
Compliance with UK Regulations
UK Government AI in Education Guidance
In April 2024, the UK Department for Education published guidance on generative AI in education. We align with their key principles:
- Teachers remain in control: All AI-generated content (marks, feedback, questions) can be reviewed and edited by teachers
- Transparency: Students and teachers know when AI is being used (clearly labeled in the interface)
- Appropriate use: AI is used for low-stakes formative feedback, not high-stakes summative assessment without teacher review
- Safeguarding: Content moderation and age-appropriate language in all AI outputs
- Data protection: Compliance with UK GDPR (see below)
UK GDPR Compliance
As a UK-based platform, we comply with UK GDPR requirements:
- Lawful basis: We process data based on contractual necessity (to provide the service) and legitimate interests
- Data minimization: We only collect and process data necessary for the service
- Purpose limitation: AI processing is limited to educational support and marking—data is not used for advertising or unrelated purposes
- Transparency: Clear information about data processing in our Privacy Policy
- User rights: Parents, teachers, and students can exercise GDPR rights (access, rectification, erasure, etc.)
- Third-party processors: We use reputable providers (OpenAI, Google Cloud) with appropriate data processing agreements
Children's Privacy
For students under 13:
- Parental consent is required
- Use must be under teacher supervision in an educational setting
- Teachers/schools are responsible for obtaining necessary consents
- We provide clear information to help schools meet their GDPR obligations
Limitations & Known Issues
We're honest about what AI can and can't do:
⚠️ AI Marking Can Make Mistakes
AI is not perfect. It may occasionally award incorrect marks, especially for nuanced or ambiguous answers. Teachers should review AI marks before finalizing grades for formal assessments.
⚠️ Students May Try to Game the System
Some students will test the boundaries of AI hints to get answers. Our guardrails help, but nothing is foolproof. Teachers should use AI features as part of formative assessment, not as a replacement for proper exam conditions.
⚠️ AI Feedback May Lack Nuance
AI-generated feedback is helpful but can't replicate the understanding and empathy of a human teacher. It's a starting point, not a replacement for teacher-student dialogue.
⚠️ Bias in AI Models
Like all AI systems, the models we use (GPT-4o-mini) may reflect biases present in their training data. We monitor for unfair outcomes and encourage teachers to report any concerns.
Guidance for Teachers
Best Practices for Using AI in GCSE Python Lab
✅ Do:
- Use AI marking for formative assessments (homework, practice) to save time
- Review AI marks for important summative assessments before releasing to students
- Enable AI hints for practice mode to support independent learning
- Disable AI features in test mode for formal assessments
- Teach students about responsible AI use and the importance of understanding over answers
- Monitor help usage analytics to identify students who may be over-reliant on AI
❌ Don't:
- Rely solely on AI marking for official GCSE mock exams without teacher review
- Assume AI feedback is always correct—verify it makes sense
- Use AI-generated questions without reviewing and editing them first
- Allow students to use AI hints as a shortcut to avoid thinking
- Submit AI-assigned marks directly to school systems without verification
Our Commitment
As the creator of GCSE Python Lab, I'm committed to:
- ✓Transparency: Being honest about how AI works, what it can and can't do, and the trade-offs involved
- ✓Teacher control: Ensuring teachers always have final say over marks, feedback, and pedagogy
- ✓Student welfare: Designing AI features that genuinely support learning, not undermine it
- ✓Compliance: Following UK regulations and best practices for AI in education
- ✓Continuous improvement: Listening to feedback, monitoring outcomes, and updating guardrails as needed
If you have concerns or suggestions about how we use AI, please get in touch. This is a conversation, not a finished product.
Questions or Concerns?
If you have questions about our AI features, guardrails, or compliance, please contact:
Forge Class Software Ltd
Email: enquiries@forgeclass.co.uk
Subject line: "AI Ethics Inquiry - GCSE Python Lab"