Date of the Workshop: Tuesday 29 July 2025
Location: University of Liverpool
Introduction
Automated grading plays an important role in helping lecturers manage large classes. It has been widely employed in courses assessing routine procedures such as differentiation, integration and Gaussian elimination. However, modern assessment packages are backed by computer algebra systems that are sufficiently powerful and flexible to mark much more advanced material. Since many UK mathematics degrees now have large classes at FHEQ levels five and six, this is an important area for development.
Workshop Overview
In-person workshop, consisting of three invited presentations (Y Bazlov, C Lawson-Perfect and G Woods) and five contributed presentations, followed by a discussion of future directions for automated grading. The workshop was attended by sixteen participants (including the organisers). Fifteen of these are members of staff from universities across the UK; the other is an employee of DigitalEd (the company responsible for the Mobius platform).
Main Themes and Ideas
- Fair marking in complex questions
The workshop included extensive discussions of strategies for fair marking using different automated grading platforms. These included step-by-step question structures, pre-emptive validation of students’ answers (e.g. checking for common typographical errors before the answer is submitted), and follow-on marking techniques. - Automated grading and artificial intelligence (AI)
The challenges posed by AI produced a lively and highly instructive debate. It was broadly agreed that creating assessments that fairly test undergraduate students whilst not being possible for AI systems to solve is extremely difficult. Strategies such as directly referring to course materials, and asking for corrections to flawed solutions, limit the capabilities of current AI implementations. However, these may turn out to be stop-gap measures as AI continues to evolve. One suggestion was to set assessments under the assumption that students will use AI, but this can only work if the output produced requires further interpretation or analysis; if AI simply solves the problem as presented then such assessments have no summative value. - Automated grading in formative assessment
This theme arose in several presentations delivered at the workshop. Several mathematics departments use automated grading for formative, rather than summative, assessment, and provide targeted support to students based on their results. Given the pressures on staff time caused by large classes, support of this type would simply not be possible without automated grading. - Automated grading and the changing assessment landscape
Several universities represented at the workshop have recently completed (or are in the process of completing) review processes, one aim of which is to reduce the number of invigilated examinations. Automated grading for assessments with
individualised questions for each student can play an important role in meeting the challenges posed by these requirements whilst maintaining academic integrity.
Emerging Implications and Recommendations
The workshop has initiated ongoing discussions between colleagues at Liverpool, Birmingham and Aston. It is evident that many institutions face similar challenges (large classes, reductions to invigilated examinations, effect of AI on assessment), and whilst automated assessments are well developed for more basic courses, there is much to be gained from using it in more advanced modules as well.
Several delegates reported that their institution is currently undergoing (or has just completed) a review process, one aim of which is to reduce the number of invigilated examinations (see Automated grading and the changing assessment landscape, above). The organisers are aware of several additional institutions to which this also applies. These reviews are likely to have a profound effect on mathematics courses across the country; a workshop on this subject is likely to be useful to many academic staff. Automated grading can play a major role in the provision of alternative assessment formats.
Resources and Links
Presentation abstracts and slides are available to download: Automated grading in mathematics & statistics: beyond the basics
Acknowledgments
The event was organised by Ian Thompson and Jessica Banks.
Presentations were delivered by:
- Yuri Bazlov (University of Manchester)
- Christian Lawson-Perfect (Newcastle University)
- Robert Leek (University of Birmingham)
- Tim Lowe (Open University)
- Jamie Mason (Durham University)
- Maciej Matuszewski (Durham University)
- Colin Steele (University of Manchester)
- Gareth Woods (Aston University).
Conclusion
This was a highly successful event which brought together experts in automated grading from a range of UK universities. Many of the ideas discussed can be employed on any platform, and there is a great deal of scope for future development and collaboration. We are grateful to the IMA for its support.
