Why might AIR be useful in terms of marking and feedback for pupils’ writing?
Analysis of existing marking practices revealed several key issues that we knew AIR could address effectively:
1. The expectations on Primary teachers to plan, deliver and assess the range of subjects on the National Curriculum is a heavy burden. The marking of writing – just one subject of many – is perhaps the most time-consuming. While live feedback during lessons is often cited to be most effective, in a school such as ours in which each classroom includes International New Arrival pupils with no English, majority EAL pupils, significant SEND needs and pupils who are achieving well and striving to meet age-related expectations, the capacity to provide such live feedback to all pupils is limited.
2. After lessons, teachers were spending excessive time on the marking and feedback of writing, leaving little room for the planning of personalised support.
3. The feedback itself wasn't always engaging pupils effectively in the writing process. Rather, it focussed heavily on transcriptional skills yet lacked the explanation of the improvements needed. Moreover, teachers have limited scope to provide bespoke feedback - celebratory and developmental - which develops pupils’ understanding of their identity as a writer.
Developing a Tailored AI solution based on these insights, we developed the formative functioning approaches of a product that was initially focused on summative assessment in Key Stage 2 (another aspect of AIR though not the focus of this Case Study). We wanted to inform the roadmap of the formative functioning and had a vision that the system would provide the following features within the classroom:
1. A shift in the feedback ratio: We implemented a 2:1 ratio of positive to constructive comments, crucial for building confidence and engagement.
2. Time-efficient marking: By streamlining the current teacher approach and utilising AI in the marking process, we aimed to free up teacher time for targeted tutoring and support, bringing the formative writing marking process down to under 5 minutes per pupil.
3. Actionable feedback: Ensuring that the feedback provides specific, manageable improvements that pupils can address independently or with support immediately after they receive it.
4. Personalised approach: Ensuring that the feedback is tailored to individual pupil needs, recognising the diverse challenges within the school population. We wanted to ensure feedback was stage, not just age, driven. There is little point in a Y5/6 child mastering the use of semi-colons when they cannot yet consistently use full stops, for example.
5. From AIR screen to pupils’ books: With a focus on quality feedback for pupils and efficiency for teachers, feedback will be printed using thermal printers and stickers which can be stuck into pupils books. AIR will be introduced in September 2024. As AI continues its rapid development and teachers strive to exploit AI to support their role, I am sure our system will see further refinements as we seek to maximise its potential.
Key Learning
Start with a clear understanding of specific challenges faced by your schools (teachers and pupils). Collaborate closely with AI experienced developers to ensure solutions are tailored to educational needs. We found that AI can be a powerful complementary tool when used alongside human expertise, enhancing rather than replacing educational best practices. Build upon and develop existing best practices rather than replacing them completely. The iterative process of refining AI-generated content highlighted the importance of continuous learning and adaptation in educational technology integration in order to create “the best of both worlds”. Ensure new tools free up teacher time for high-impact activities like personalised support. Prioritise solutions that are inclusive and adaptable to diverse pupils' needs.
Risks
Over-reliance: Teachers could use the AI to provide the feedback without evaluating it which could potentially diminish the role of human judgment and intuition in the assessment process.
Human Agency: There is a clear need to maintain human control in order to harness the potential of AI while maintaining the integrity and human-centeredness of the educational process.
AI Inaccuracies: AI could hallucinate or contain errors.
GDPR: As with any technological integration in education, there are valid concerns about data privacy and security. It is imperative to establish robust protocols for protecting pupil information and ensuring transparent use of AI in educational settings.