The focus of this blog is in two parts: (a) consideration of making effective choices of when, and if, we should use digital approaches to assessment in the classroom, and (b) to clarify some of the current misunderstandings about Artificial Intelligence (AI) and our learning futures.

Assessment is central to learning and it’s important to remind ourselves of this when planning in our roles as teachers. It is a common mistake to think of assessment in terms of tests, grades and levels, but if used well, teachers know that assessment provides a rich, detailed understanding of individuals and how they are developing as learners. New digital technologies can support assessment in a range of ways, but it is important to be clear that data doesn’t drive your professional decision making, it should be there to support what you do because you are the experts.

Ongoing research from the Education Endowment Foundation says that a balance is necessary between efficiency and educational effectiveness. Essentially, this means a consideration of how to create opportunities to use digital technologies to collect information (think online polls, word cloud development etc., using laptops/tablets). The outcomes of these reflective tasks provide an immediate 'view of the room' and they serve as a springboard to initiate further learning or ideas from pupils. Such approaches can also be viewed as more inclusive as they allow less confident pupils to provide input anonymously and research suggests this significantly boosts engagement and motivation to learn. In terms of creating summative assessments of student achievement, digital systems can save marking time (automated marking), these can be linked to student records (assessing progress over time) and, where it’s useful, the data can be part of the evidence drawn together to share with pupils and their parents when reviewing their achievements and identifying challenges related to learning during each term/school year. But this all comes with an important caveat of remembering that progress in education is rarely linear and no amount of efficient modelling of data or use of online resources in class can change the individual nature of how each pupil learns and how each teacher teaches.

The EEF studies and other research in primary education emphasise the point that digital assessments are complementary to established learning practices, rather than a replacement or alternative. It is perhaps most important to remember that in a fast-changing educational landscape the narrative should not be about constant replacement, but instead to know what works and how teachers can boost their personal pedagogical toolkits in a broad range of ways.

Two words that have been hard to avoid over the past year in education have been Artificial Intelligence or AI. In particular there has been something of a moral panic around the launch of the Large Language Models (LLMs) such as ChatGPT and CTRL: these generative models can classify and generate text, answer questions and translate text from one language to another.  

These AI models have been met with excitement, fear and some scepticism, raising questions such as: 

• Will children no longer need to learn to write?
• Are all our assessments now using AI?
• Will the computers take over from teachers and put them out of work?
• Will children use AI to help them cheat in their assessments?

My answer to all of these echoes my earlier quote from Dylan, "maybe, in certain conditions". To try and reduce the potentially negative effects that such questions propose, we should be clear about what AI is and what it does in terms of assessment. There are, at present, very few standardised assessments that use AI and certainly none that are relevant to a primary classroom. We might use computer-generated assessments (created by the teacher on a computer and taken by pupils on a computer), or we might have assessments that can be computer marked (e.g. a multiple choice tests), but none of these include AI because they don’t need to. It is important that these differences are clearly communicated by schools.

Where I expect AI might have some impact on primary classrooms relates to the use of LLMs and creation of homework (this could be written or creative responses to a topic). Pupils are already becoming adept at using AI models to provide for example, written answers to a question, to research a subject or to create an image. Such use of these models should not be a shock to us, but we do need to address how pupils might use the content generated, and within schools, it is vital to communicate clearly what is expected in terms of individual input for the purposes of assessment. This could in fact be an opportunity to broaden assessment practice – the core of any assessment could become a presentation, or a discussion activity where pupils have to explain their work and their peers can become part of the assessment process. Doing this not only enhances how pupils have to evidence their learning but it is better assessment – it tells a teacher more about the extent to which learning has happened and where an individual needs to go next.

There are practical implications for all of the ideas that have been discussed here: good quality and reliable digital technologies rely on excellent access to high-speed internet and, particularly in schools, robust safety measures for working safely on screen and online. The level of investment required is significant, but the future is now and if we want our digital natives to make the best use of new technologies as they unfold, we need to be able to support them in every way.

Mary Richardson is Professor of Educational Assessment at the IOE: UCL's Faculty of Education and Society. She is currently the leader of the EdD Professional Doctorate programme and a tutor on the MA Educational Assessment. Her doctoral candidates focus on a wide range of topics related to assessment including: social media representation, ethics and anxiety in testing, and comparative judgement, to name just a few. Mary's own research is anchored in assessment with specialist expertise in student experiences, public understanding of testing, the ethics of assessment and currently, what AI might mean to all our assessment futures. She sits on the Research Board for Qualifications Wales, for AQA Exam Board's Research Department and is the co-convenor of the British Educational Research Association's Special Interest Group (SIG) for Curriculum and Assessment. When she is not thinking about assessment, Mary is a keen middle distance runner and passionate about northern soul music.