It is a rite of passage for students reaching the final stages of secondary education, a summation of years of hard work to secure a place at university or college. But this year, A-level results day has felt very different.
With coronavirus shutting schools for the vast majority of pupils up and down the country, exam season was thrown out too. How should students be judged?
With algorithms, apparently. Roughly 300,000 students who received results on Thursday will almost all have had a grade assigned to them based on a piece of code put together by Ofqual, the exams regulator. But not everyone will be satisfied.
Early figures suggest 36pc of results in England were marked down by one grade, while 3pc were down by two. Ofqual numbers also show a 4.7pc rise in A and A* grades in private schools versus 2019, and just a 0.3pc shift in the needle for state institutions.
“These outcomes are not surprising - it is well established that the use of algorithms to score and predict creates risks that existing social biases and inequalities will become exacerbated and entrenched,” says Carly Kind, director of the Ada Lovelace Institute.
Years of brushing up on integration calculations, Chaucer, hydrocarbons or game theory, then, were for nought, as students rage that their results could have gone differently had their teachers’ predictions on performance been given greater weight versus statistical modelling done by a computer.
The opportunities that have slipped away for young people as a result of coronavirus has already led them to be deemed the “lost generation”. Has a drastic mistake been made by putting their future in the hands of artificial intelligence?
For one, little transparency has been afforded to date on the algorithm. The government is expected to release a near-150 page document that could shine some light on its mechanism, but that is yet to emerge.
n the run up to results day, Ofqual made a number of changes to how it said it would assess students. First, it asked teachers to submit predicted grades, along with a ranking of students in those classes. But later, it decided predicted grades would play a minimal role in determining grades versus statistical modelling for classes of over 15 students.
In many schools, then, it meant disaster was waiting to happen. Curtis Parfitt-Ford, an A-level student at Elthorne Park High School in West London who received results on Thursday, saw tears on the faces of friends who had been marked down by the code.
One student he knows was marked down “two entire grades”, snatching their university place out of their grasp and forcing them to find a new offer through clearing. “When we found out it was all being done by computer... that really hit as a shock,” he says.
It’s why Parfitt-Ford is bringing a challenge to the government over the algorithm with the support of Foxglove, a team of legal specialists challenging “digital injustices”.
Among the concerns raised by the organisation is the fact that an individual student’s life chances “hang on an estimated balance on their school’s historic performance”, given the data fed to the algorithm for classes with more than 15 students.
For Cori Crider, lawyer and co-founder of Foxglove, it means the algorithm robs students of an assessment that takes them on their own merit, particularly in state schools where classes tend to be much bigger, turning code into bias that exacerbates existing chasms in Britain’s education system.
“We think that's legally super problematic, the whole point of Ofqual is to assess individual achievement. If you are substituting teacher's assessments with this algorithm in the name of trying to maintain your bell curve then you're not doing what the statute says you can do,” she says.
Foxglove is also arguing that the algorithm violates a key part of GDPR and the UK Data Protection Act, which “provide significant protections from automated decisions” that may have significant consequences for people.
With A-level results determining whether or not a student secures a spot on their chosen university course to pursue a dream career, the algorithm must be held to account if it has been unfair. Experts have warned that algorithms are easily skewed by data unless properly assessed and can proliferate existing biases.
They have also been surprised by a lack of external audits or inspections to establish if the algorithm was up to task. The Home Office has previously had an algorithm used to grade visa applicants to the UK scrapped after being described as “racist” by campaigners.
“We won the UK’s first judicial review of a government algorithm and this one looks equally unfair and biased to us,” Crider says.
Gavin Williamson, Education Secretary, claimed the “majority of young people will have received a calculated grade today” that will help them on towards “the destination they deserve”, stressing the option to appeal based on any disparities with mock exam results.
Parfitt-Ford is hesitant to jump the gun in dismissing the Education Secretary’s claims until statistics are available, but he says that it is “frankly disrespectful” for a minority of students, however small the group may be, to be unfairly treated.
“The fact that we're in a bad situation, the fact that things are difficult doesn't mean we should go for the worst possible solution,” he says. “We need something better than this.”