the e-Assessment Association

Efficiency and Evolution: How AI can transform the EPA landscape in 2024

Efficiency and Evolution: How AI can transform the EPA landscape in 2024

A blog by Tad Chapman, Vocational Education Solutions Lead, at risr/

The longstanding challenge for the end-point assessment (EPA) sector has been reconciling the rising demand for assessments with a limited ability to scale in parallel – leading to strain and pressure on the EPA system.

So, as we look ahead to what 2024 will bring, I expect this need for greater efficiency will remain a key issue for the sector to try and address.

Ofqual’s latest data on the recent growth in the number of apprenticeship completions highlights that from 2020/21 to 2022/23, the number of EPAs completed increased from 13,405 to over 110,000.

To compound matters further, the next 12 months will also see a number of reviews to existing apprenticeship standards, which could bring significant change to assessment plans. This could result in huge operational impact for end-point assessment organisations (EPAOs) in terms of how these changes are delivered.

So, if EPAOs are to stay ahead of the game in 2024, it’s crucial that they remain cognisant of ‘emerging’ technologies, ones that can make a marked improvement to the efficient delivery of end-point assessments.

Artificial Intelligence (AI) is one huge area of opportunity here. Within vocational education, discussions on AI have largely focussed on concern around the prevention of cheating, with a small acknowledgement that AI may be able to primarily assist training providers, rather than EPAOs. But I believe there’s a vast range of other useful applications that are yet to be explored for EPAOs.

A significant boost to efficiency that would be impossible to ignore, would be the use of Large Language Models (LLMs) to generate question banks based on inputted occupational standards. These LLMs could be embedded within the question authoring system, providing the ability for the requisite assessment information to be fed in without user intervention – in a similar way to how APIs work. As a result, this could save EPAOs significant time and cost on question writers, with experts able to refine questions, rather than create them.

Another huge efficiency boost for the EPA sector would be the exploration of AI for boosting the productivity of administrators. If this came to fruition, we could see a huge reduction in the cost and logistical planning hurdles that are often associated with EPA. For instance, AI could be used to coordinate diaries of the candidate and assessor via an online booking platform, meaning candidates could book in and take their exam sooner.

Oral assessments represent another area that can benefit from an infusion of AI. As speech-to-text technology improves, the recording of the assessment could be directly inputted into an LLM, transcribed, and then compared against a marking rubric. It means a live human assessor could have a first draft of a mark done almost immediately and wouldn’t need to spend hours revisiting transcripts or recordings.

Of course, it still needs to be checked by human intelligence, but this would be significantly more efficient and would increase the capacity of an assessor to deliver more assessments, as well as providing more standardised, consistent assessment decisions.

These three AI developments are just the tip of the iceberg when it comes to possible improvements that AI can make to the operations of EPAOs.

Critically, this isn’t to say AI is the panacea to the issues the EPA sector is facing: human intelligence remains the key driver to success. Rather, it’s to point out that we, as an industry, have the ability to drastically improve the efficiency of how we work, and AI could be an incredibly valuable ‘co-pilot’ in providing solutions to the widespread issues facing the EPA sector.

The common issue I’ve come across to date is that EPAs perhaps don’t have the resource to fully explore where they can utilise AI, which brings us back to the overarching issue. Resolving challenges in the EPA space requires continuous innovation and development. In 2024 and beyond, EdTech solutions providers must continue to prioritise solutions that deliver measurable and tangible impact for the sector.

risr/ is a global assessment software company with a decade of experience delivering high-stakes assessments.

Share this: