Algorithmic bias: should students pay the price?

  • PDF / 375,338 Bytes
  • 2 Pages / 595.276 x 790.866 pts Page_size
  • 27 Downloads / 173 Views

DOWNLOAD

REPORT


CURMUDGEON CORNER

Algorithmic bias: should students pay the price? Helen Smith1  Received: 15 August 2020 / Accepted: 18 August 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020

Thanks to the COVID-19 pandemic, rather than the usual exams, A-Level students in the UK faced the novel process of being awarded grades generated by calculation determined by an algorithm (Coughlan 2020b) as administered by the Office of Qualifications and Examinations Regulation (Ofqual 2020). One need to only explore #Alevels via Twitter to find cases of students who have been tremendously downgraded by this process; in one incidence from a predicted A grade to a U (@TykeTaffy 2020). The grades awarded understandably generated uproar and resulted in the destabilisation of many members of this year’s student cohort who depended on their results to move on with their lives to further education, training, or employment. The entire fiasco had been due to the application of an algorithm which goes against the United Kingdom’s (UK’s) Government’s very own recently drafted guidelines: The Office for Artificial Intelligence’s (2020 p. 36) guidance on the use of AI in the public sector stipulates that there must be “a process for monitoring unexpected or biased outputs”. Ofqual (2020) stated that they “performed an equalities analysis of calculated grades” upon their model which determined this year’s grades; they claimed their analysis showed “no evidence that this year’s process of awarding grades has introduced bias.” However, this was not found to be the case; the distribution of downgrading had been noted to favour students attending private/independent schools (Lee 2020) whilst students from disadvantaged backgrounds had been hit hardest (Adam and McIntyre 2020). Ultimately, the algorithm used to restore grading balance and determine student’s final grades was both inconsistent and unfair (Coughlan 2020b). Public pressure resulted in the first U-turn by the Scottish Government whereby grades would be re-issued and based solely on teacher judgment (Coughlan 2020a). Had it not

* Helen Smith [email protected] 1



Centre for Ethics in Medicine, University of Bristol, Bristol, UK

been for peer pressure applied by the devolved governments to follow suit, the UK Government might not have decided to finally abandon the contested algorithmically issued grades. Without this change of heart, many student’s futures would have been sacrificed in the name of maintaining a standard, which had not even been uniformly applied across the UK’s four nations, England, Scotland, Wales and Northern Ireland. Every nation has had their education system impacted by the pandemic, there has been no globally coordinated response for managing exam results to ensure fairness worldwide. Yet, whilst the UK Government concerned itself with devaluation of grades, some other nations simply accepted an overall jump in grades. For example, France utilised results available from exams taken earlier in the student’s courses; they enjoyed record pass r