Ofqual (England’s Office of Qualifications and Examinations Regulation) has issued an update on its approach to regulating the use of AI in the assessment sector. By doing so, it acknowledges that the discussion has moved on from whether AI has a place in assessment. Awarding bodies and regulators must now work to implement governance frameworks, collect evidence and build the trust necessary to introduce AI-enabled innovations sensibly and at scale.
Ofqual’s recognition that the speed of AI-adoption in assessment design, delivery and participation requires review of their approach is a positive development. Its commitment to an evidence-led approach avoids knee-jerk reactions and provides confidence for a sector keen to innovate.
Through its regulatory approach Ofqual aims to support innovation and growth in the sector, where three elements are relevant:
Ofqual’s remit is limited to England. There could be as many different approaches to regulating AI in assessment as there are individual regulators around the world. However, the number of people attending the presentation given by Michael Hanton, Ofqual’s Deputy Chief Regulator at the recent e-Assessment Association International Conference suggests that some regard it as an example to follow.
Not everyone shares that view. Nevertheless, candidate behaviour and wider societal expectations require regulators to acknowledge the impact of disruptive technologies and adapt their approach. Effective regulators will also grasp the opportunity to steer developments in a positive direction.
RM is adding AI-marking capabilities to our assessment platform. Through proof-of-concept work with existing customers we have established how AI-enabled marking benefits awarding organisations and candidates. RM’s approach involves compliance with local and international regulations to incorporate AI alongside human involvement.
Our experience of talking to organisations globally confirms that enthusiasm for AI marking in some parts of the world than in others. Some countries are looking to embrace AI-enabled marking more quickly than would appear to be possible under Ofqual’s requirements. Rapid innovation obliges awarding organisations and assessment technology providers to balance potentially competing demands. These can include regulatory concerns about candidate data as well as the protection of their own and others’ intellectual property.
RM’s remote proctoring and invigilation capabilities include AI elements which can be deployed where appropriate to meet an awarding body’s needs. For now at least it appears that AI-only invigilation or remote proctoring is incompatible with Ofqual’s requirements. RM provides awarding bodies with the ability to deploy an AI agent to monitor and intervene as necessary, reinforced by a clear process to mobilise human intervention.
Ofqual acknowledges the problematic nature of students using generative AI in non-exam assessment. Anecdotal evidence suggests that not all students use the generative AI shortcut for coursework. However, finding examples of students failing to resist temptation is not difficult, especially in higher education, while institutions’ reactions can be inconsistent.
Placing the onus on awarding organisations to manage the risks associated with AI misuse and ensure the validity of student work boosts the need to develop reliable and scalable tools to detect AI use. An alternative approach could be to create frameworks that acknowledge AI use and appreciate how the view of the piece being assessed changes as a result.
As a responsible partner to awarding organisations, RM will apply its values of balancing innovation with appropriate governance and risk management. Doing so not only serves our own interests but also those of our awarding body customers who must maintain integrity and trust in their qualifications. But most importantly it’s in the interests of the people we all serve - the candidates sitting assessments that can determine their life chances.