British, International & IB Curriculum AI Lab – A Formative Assessment Test Bed

Welcome to the British, International & IB Curriculum AI Lab!

Most AI marking systems cannot clearly explain why a student received a grade. However, this one can.

We are building an AI marking test bed for British,GCSE, IGCSE and IB classrooms on three foundations: 

• Grounded AI anchored in board-specific mark schemes and model answers
• Explainable AI that traces every mark to exact rubric criteria with clear reasoning paths
• Safe AI where teacher approval is a non-negotiable checkpoint.

Every decision is traceable, evidence-backed and auditable, turning marking from a black box into a system schools can fully interrogate and trust. The result is consistent, high-precision feedback at scale without removing professional judgement.

Built in line with examiner standards and evolving regulatory expectations, including Ofqual, GDPR, and the EU AI Act, the Lab operates as a controlled, real-world environment for reliable AI in assessment.

If you want to understand how it works, and how your school can help shape the future of AI in education, keep reading.

What 100+ teachers told us - and what the research confirmed.

When we spoke to teachers across IB, IGCSE & British Independent schools, the same pattern emerged regardless of curriculum, school size or geography. 

Students preparing for high-stakes examinations - GCSEs, A Levels, IGCSE, the IB Diploma Programme, and 11+ entrance - require frequent, highly accurate, criterion-aligned feedback at every stage of practice.

But the volume of practice needed to move a student from one grade boundary to the next far; exceeds what a teacher can realistically mark within a working week.

This is not a problem of motivation. It is not a gap in teaching quality. It is a structural constraint on time.

And it is precisely this constraint that the AI Lab is designed to solve.

A live test bed for curriculum-aligned AI marking - built to examiner standard, grounded in verified sources.

The Lab is a working research and development environment where AI-powered formative assessment is continuously built, tested, and refined against real marking standards from British, IGCSE, and IB curriculum boards.

It is designed in line with Ofqual’s principles for AI in assessment and co-developed with subject experts across all three frameworks.

At its core is a teacher-led human-in-loop system powered by an AI agent infrastructure (“The Brain”), built on a Small Language Model (SLM) architecture to deliver high accuracy outputs.

Grounding Layer -  sources the system is trained on:

• IBDP, IGCSE and British board-specific official mark schemes & rubrics, Official Textbooks retrieved via RAG
• Chain-of-thought prompting is used to guide the AI Agents to work together and execute their actions
• Human written model answers, authored by practising Oxbridge-educated teachers
• Past and predicted papers from the curriculum data bank
• Reasoning traces: Human written, detailed, stepwise answer explanations for every response

Human in the loop, by design:

Teachers remain central to the system:

• AI completes first-pass marking against the exact rubric
• Teachers review, edit, and approve
• No feedback reaches a student without human validation

Curriculum-aligned outputs

The system produces:

• Evidence-based feedback aligned to mark schemes and rubrics
• Fully cited responses (syllabus, textbook, examiner criteria)
• Zero-hallucination outputs - every claim is traceable to a verified academic source
• Personalised remedial pathways that close the loop: attempt → feedback → adjustment → reattempt

A connected digital ecosystem

Built to scale across schools while remaining interoperable and inclusive by design.

Aligned with:

• Ofqual guidance on AI in assessment
• UK AI Safety Institute principles
• IB Digital Blueprint

What schools will experience

Schools joining the Lab will see, in practice:

• How grounded AI marking behaves under real exam conditions
• How adaptive feedback loops improve student outcomes over time
• How the system handles edge cases in high-stakes marking

This is explainable, accountable AI - built for the realities of the classroom, not just the promise of technology.

Before the first line of code was written, three principles were established as non-negotiable.

Transparency

The “black box” problem - the inability to explain how an AI assigns a grade - is the single greatest barrier to trust in AI-led assessment.

This Lab is designed to eliminate that opacity. Every output is fully traceable:

• the exact rubric criterion applied
• the specific mark scheme reference used
• the reasoning path linking the student response to that judgement

The result is not just a mark, but an auditable decision.

Authenticity

Academic integrity in the AI era is not hypothetical - it is immediate. The system is grounded in a structured evidence layer:

• board-specific mark schemes
• curriculum-aligned textbooks retrieved via RAG
• human-written model answers
• controlled reasoning frameworks

These are not loosely combined. They act as constraints on generation. Outputs are therefore:

• curriculum-aligned
• citation-backed
• resistant to hallucination

Integrity

AI in education operates within tightening regulatory boundaries. This Lab is built in alignment with AI Policies and Governance:

• EU AI Act principles
• Ofqual guidance on AI in marking
• IB digital assessment frameworks

Human judgement is not an override. It is part of the system architecture.No feedback is released without teacher approval. Teachers retain control at every stage:

• editing prompts
• adjusting marks
• approving or withholding outputs

The system does not replace professional judgement. It enforces it.

The system is built around a single objective: reduce teacher workload while delivering faster, more precise feedback to students.

For teachers

Live dashboards with cohort-level visibility
Class Performance Insight tracks attainment, progress and distribution across classes in real time, giving teachers a clear view of what is holding cohort performance back.

Assessment creation in minutes, not hours
Generate exam-style papers directly from a vetted curriculum databank - fully aligned to board mark schemes and free from AI hallucination.

Full control over AI outputs
Every stage is editable. Teachers can adjust prompts, override marks, and refine feedback before anything reaches a student.

Granular, skill-level gap analysis
Pinpoint exactly where a student is losing marks - by topic, skill, and rubric criterion - across every subject.

Targeted remediation, generated instantly
Auto-create follow-up questions, model answers and practice sets mapped precisely to the student’s areas of weakness.

Global benchmarking via a universal leaderboard
Compare performance across British, IGCSE and IB cohorts worldwide, with normalised metrics.

Measured efficiency gains
Up to 90% reduction in marking time, with paper creation effectively automated.

For students

Immediate, criterion-linked feedback
Every response is marked against the exact rubric, with clear justification for each mark awarded - including extended and subjective answers.

Stepwise, examiner-standard model answers
Structured solutions with method breakdowns and guided hints - written and reviewed to examiner standard.

“Teach Me” mode with video solution walkthroughs
On-demand explanations for every question, designed to replicate one-to-one teaching.

Personalised study and remedial action plans
Built from performance data so students can adjust, improve and try again. Updated continuously after each attempt.

Precision over generic feedback
Weak areas are explicitly tied to mark scheme criteria, so students know exactly what to fix and how.

The Lab includes a specialised AI Governance and Literacy course developed specifically for teachers and curriculum leads in British, IGCSE and IB schools. 

It is built around the real questions teachers are already asking - and the critical ones they have not yet encountered.

Seminar Topics:

School AI Readiness
Preparing teachers for the realities of AI in the classroom - including a practical Data Privacy Impact Assessment (DPIA) framework tailored to AI-supported assessment tools.

The Irreplaceable Teacher
Why professional judgement remains central to high-stakes assessment, and how AI should be positioned as augmentation, not replacement.

Curriculum Updates (IB, IGCSE, British)
Key changes across major boards, and what they mean for assessment design, feedback cycles and classroom practice.

AI Policy & Governance
Clear, actionable frameworks for safe adoption - aligned with Ofqual guidance and emerging UK AI Safety Institute benchmarks.

The Black Box Problem
Understanding why opacity undermines trust, and how explainability changes the way AI can be used in assessment.

Hands-on lab sessions:

Explainable AI in Practice
Verifying that AI output is curriculum, rubric and mark scheme aligned, and awards the correct marks. Participants work with live reasoning traces.

Safe AI Tutoring
Hands-on practice with tools co-created with teachers, authenticated for transparency, fairness and accountability

Human-in-the-Loop Systems
Maintain meaningful oversight in marking and feedback. Learn where intervention matters - and where it doesn’t.

Hallucination Detection
Identify, test and correct AI-generated errors before they reach students, using structured validation approaches.

Bias & Risk Management
Recognise where bias enters AI-supported assessment, and apply practical safeguards to mitigate it.

Outcome

The goal is not just to gain technical expertise. It is to imbibe informed control.

Teachers leave with a clear understanding of what the system is doing, where it is reliable, where it is not - and the confidence to make professional decisions accordingly.

Built on eighteen years of working at the standard your students are assessed against.

PiAcademy has worked with over 20,000 students and 200+ schools across the UK and internationally. The tutors and model answer authors whose work sits inside the grounding layer of this platform are Oxbridge-educated, practising educators in leading British independent and international schools - working daily with the same mark schemes your students are assessed on.

Our work has been recognised through national innovation programmes, including the Innovate UK Fast Start Innovation Award and the Unlocking Potential Award. The AI Lab itself has been developed in alignment with emerging 2026 Ofqual guidance on AI use in assessment, ensuring that every output meets the level of traceability, reliability and examiner consistency expected in high-stakes environments.

The Lab exists for one reason.

After eighteen years working at the sharp end of British and international curriculum assessment, one constraint became impossible to ignore: student outcomes are shaped not by the quality of teaching, but by the speed, precision and consistency of feedback - and that system is limited by human marking capacity.

This is a structural problem. And it requires a structural solution.

The AI Lab is our response: a controlled, curriculum-aligned environment where marking can scale without compromising examiner standards.

We are now inviting a small cohort of schools to work with us at this stage - not as users, but as partners helping define what responsible, transparent AI marking should look like in practice.

A select group of forward-thinking schools across British, IGCSE and IB curricula is being invited to join the first phase of the Lab.

Participating schools will work directly with subject matter experts, practising examiners and the PiAcademy team to test, evaluate and refine AI-powered formative assessment systems built to examiner standard.

Register your school’s interest today! We will be in touch within two working days.