It has long been clear that GCSE results and other standardised forms of testing are not an accurate reflection of intelligence. Take the case of Maksym (not his real name) the 16-year-old son of a local Ukrainian lady whom I had been helping with her English. Maksym and I had shared brief conversations about GCSEs, sixth form, university and various career paths. The depth of his questions and nuance in his analysis of various career paths were far in advance of what I witnessed at my top Russell Group university, let alone coming from a 16-year-old boy who moved here as a refugee with limited English only a few years ago.
Needless to say, I was mortified to hear that Maksym had mostly scored 4s and 5s (‘C’s in old money) across his GCSEs. Evidently, his academic performance was not reflective of his actual intelligence.
Maksym is not alone. Contemporary methods of academic assessment are rapidly being made redundant. The problem of Large Language Models (LLMs) like ChatGPT producing traditional essays and coursework are well documented. However, AI can also liberate students from the standardised, box-ticking exercise of modern mass education by opening up more personalised and verbal means of testing, allowing bright students like Maksym to get the grades they deserve.
Before the industrial revolution, education was only available to a very select few through the aristocracy and the clergy. Due to this uneven distribution of wealth, the lucky few who could pursue education received highly personalised teaching with a handful of students to a professor and exams conducted verbally. This allowed professors to assess as interlocutors, probing the depths and nuances of a student’s knowledge. With the industrial revolution came the ability to educate en masse. However, there simply was not the time nor resources available to offer such tailored education on this scale, at least not in the publicly funded schools. It is no coincidence that this kind of personalised approach to learning is a major draw of many private schools.
As a result of this reality, there was a great rise of standardisation within the education system. Tests were not verbal debates to probe logic and argumentation, but standardised “box-ticking exercises” to test how students meet certain easily and quickly measurable criteria. As many of us know from personal experience, being able to pass tests effectively often has less to do with critical thinking and academic excellence and more with gaming the system, regurgitating whatever the assessment criteria demand. As a result, actual engagement and analysis can often be penalised if it does not fit the accepted list of facts or content on the assessment criteria, particularly at GCSE level. At my state comprehensive, we were frequently reminded by our teachers that our GCSE papers, which we had spent years training for, would often be marked by a worn-out examiner on a Tuesday evening over a glass of red wine. It was therefore important to make it “easy for them to mark:” give them what they want to see and avoid any personalised nuance or style.
Fortunately, recent developments in AI can finally offer an end to this outdated system. The technological and economic reality that forced rigid standardisation on countless generations of students no longer exists. We can now return to the pre-industrial style verbal assessments, recorded in modern exam settings to avoid cheating. This will allow students to freely express their own ideas and critical analysis to flow in verbal prose, with an AI interlocutor probing their knowledge and engagement with the content. It is no coincidence that this type of assessment – a viva – is still carried out at the PhD level where the resources are available and stakes deemed necessary.
Furthermore, due to the impartiality of AI, the actual content of the students’ argumentation could be assessed in isolation. It will avoid the countless human errors frequent in the current mass marking system such as misinterpretation of handwriting, fatigue or an assessor’s lack of knowledge outside the assessment criteria. This last error in particular frequently results in the penalisation of students who think outside the box and do independent research.
Better still, the AI assessors would also be able to beat the traditional interpersonal biases of human-based verbal assessments. AI can see through the pseudo-intellectual waffle favoured by parts of the upper classes, desperate to pass themselves off as more intelligent than they actually are. Nor will an AI assessor have any personal relationships with the students that can alter the perception of their argumentation.
Educators have long talked about the need to teach and assess critical thinking within schools. Progress has certainly been made in this regard over the past generation with more emphasis on source analysis and context. However, the whole framework of our assessment system is a product of the industrial revolution. In a digital age, we must create the systems which allow us to use AI as a tool to tackle some of society’s most entrenched issues. The ability of AI to be both standardised and impartial in its application of assessment criteria, paired with its flexibility in probing students’ arguments, allows it to address regional and economic educational inequalities.
Grades should reflect a student’s work and intelligence, not their ability to game a system. Used correctly, AI can allow those bright students like Maksym, who may lack formal assessment training, to get the grades they deserve.
Tomas Mills is a Member of Bright Blue. The views expressed in this article do not necessarily reflect the views of Bright Blue.