Excelsoft, AQA Partners to Advance Ethical AI in Exam Marking

Excelsoft Technologies (“Excelsoft”), a global provider of assessment and learning solutions, and AQA, a leading UK-based exam board, today announced the creation of a joint AI taskforce to collaborate on research and development projects focused on the ethical and secure use of AI in the e-marking of handwritten student responses. For the foreseeable future, marking exams will remain a human process.

This joint AI taskforce will combine AQA’s deep expertise in high-stakes assessment with Excelsoft’s advanced AI, data and e-assessment capabilities. The collaboration will focus on designing and testing responsible AI models and workflows that support human examiners, improve consistency, and enhance the reliability of marking, while maintaining the highest standards of fairness, transparency and data protection.

“Assessment is about trust – trust that every student’s work is judged fairly, consistently and securely,” said Sudhanva Dhananjaya, Chairman and Managing Director, Excelsoft. “By joining forces with AQA, we are creating a dedicated AI taskforce that will explore how AI can responsibly support the marking of handwritten scripts, while keeping human judgment, ethics and student privacy at the centre.”

The joint AI taskforce will work on a series of co-designed research projects, including:

• Exploring human-in-the-loop AI models to support examiners in marking handwritten responses

• Developing and validating bias-aware approaches to ensure that AI-supported marking is fair to all student groups

• Designing secure data pipelines and infrastructure for handling handwritten scripts, annotations and marking data

• Creating transparent reporting and explainability frameworks so that AI involvement in marking can be clearly understood and governed

• Establishing ethical guidelines and governance frameworks tailored to high-stakes assessment contexts

“We see AI as a powerful tool to support, not replace, our expert examiners,” said Justin Coombs, AQA Chief Assessment Technology Officer. “This partnership with Excelsoft allows us to rigorously research, test and shape how AI can be used safely and ethically when examiners are marking handwritten responses, so that students, schools and regulators can have full confidence in the process.”

The joint AI taskforce will comprise data scientists, psychometricians, AI engineers, assessment designers, examiners and ethicists from both organisations. It will operate under a shared governance structure, aligned with relevant regulatory, data protection and security standards, and will engage with stakeholders – including teachers, students and policymakers – as part of its research.

“Our goal is not just to build technology, but to build trustworthy AI practices for assessment,” added Adarsh Sudhindra, Excelsoft’s Vice President – Products & Innovation. “This collaboration aims to set benchmarks for how AI should be used in marking – always accountable, always explainable, and always in the best interests of learners.”

Insights and findings from the joint research will be used to inform future product roadmaps, marking services and policy discussions, and may be shared through jointly authored white papers, pilots and industry forums, subject to appropriate governance and approvals.