With the advent of advanced technologies like generative AI, it’s easy to prioritise adding them into services while forgetting the value of human involvement.  Earlier this year, Ofqual emphasised the integral role of humans when they introduced new regulations regarding AI in remote assessments.  These regulations outlined that assessment providers need “appropriate human involvement and safeguards” in high-stakes assessments due to the “high bar required to ensure that all remote assessment evidence can be trusted.”  

In short, using AI on its own cannot guarantee the same integrity as using a human invigilator.   

A human-first approach is what is necessary to create trustworthy, reliable, and robust systems.  This is something we have always prioritised at VICTVS throughout our processes, and which is only made possible through our head office staff and VICTVS Global Network.  

The Cost of Relying on Technology  

As noted by Ofqual’s recent regulatory changes, relying predominantly on technology to invigilate remote exams can put a qualifications integrity at risk.  Humans are simply more reliable than AI when it comes to accurately detecting suspicious behaviour. 

Researchers at Cambridge Judge Business School state that while using AI can have a clear advantage in data-driven exercises, the technology still falls far short when it comes to complex social tasks and ethical dilemmas.  In the words of researchers Hamza, Kamal and Shaz: “AI can rapidly learn and iterate in a controlled environment, making it less ideal for coping with highly disruptive events that require human intuition and foresight.”  AI relies on “historical data rather than intuition”, meaning it can make false assumptions about an action based on previous data inputs. 

As such, AI-driven systems are also more likely to incorrectly flag neurodivergent candidates as exhibiting suspicious behaviour.  For example, more frequent eye and head movements that are commonly associated with ADHD might be identified as a sign of misconduct.  Research from earlier this year concluded that “AI can improve accessibility, but detection software produces false positives disproportionately affecting neurodivergent students.”  

When it comes to false positives, even a small percentage can have a huge impact.  The National Centre for AI notes that even a rate of 1% can still result in thousands of false positives annually when candidates are taking multiple assessments per module, resulting in a substantial administrative burden for institutions as well as damage to student trust and well-being.  Ultimately, students flagged incorrectly for misconduct may fail their exams, be banned from future assessments, and lose out on career opportunities.  The stakes are simply too high for even a 1% false positive rate not to matter. 

The Value of Human-First Invigilation at Scale 

Instead of relying solely on AI-driven remote invigilation, combining human proctoring with modern technology ensures that human ethical reasoning and judgement remains at the forefront of our exams, resulting in greater accuracy in identifying misconduct.  As stated above, it’s clear that human invigilators have a degree of emotional and contextual intelligence, as well as intuition, that an AI-moderated exam simply cannot compete with.  

On top of the ability to more accurately identify attempts at malpractice, human-invigilated exams have other major benefits.  Our VICTVS Global Network of invigilators and exam staff is based all over the world, meaning that for the vast majority of assessments, we can match exam candidates with invigilators who speak their native language and understand their local culture.  This can help put the candidate at ease and create a more comfortable and personalised exam experience which an AI service cannot compete with. 

Candidates who believe they are being monitored by an attentive invigilator are also less likely to cheat.  Researchers at the University of Zambia concluded that “knowing an invigilator was attentive made cheating too risky, thereby encouraging compliance with examination regulations.”  Therefore, an attentive person monitoring an exam is more likely to stop a candidate cheating than an invisible algorithm. 

Human Support Around the Clock 

As well as human invigilation, both candidates and invigilators benefit from human support on exam day when any issues arise.  This can include technical or WiFi issues, as well as problems with third-party software. 

While customer support is often left to AI chatbots, we combine our technical systems with 24/7, 365-day human support to ensure candidates are always able to reach a staff member who can help them.  To make this happen, we operate support services from our head office in Leeds, as well as Colombia, the Philippines, and Malaysia.  This means that there is always someone ready and available to help, no matter the time of day. 

Technically-Powered but Human-First  

As new technologies emerge, it would be a mistake to miss out on the opportunities they present.  When it can improve vigilance, data analysis, and monitoring capabilities, AI should absolutely be used in assessments.  But that doesn’t mean it should replace existing, trusted methods like human invigilation.  Not only is human invigilation a trusted and reliable mode of monitoring that leads to greater accuracy in identifying malpractice, but it can help put candidate at ease and create a better exam experience.  Supporting this human-first approach with tried and tested technology is the best way to ensure a robust, reliable service for your candidates.