THIS IS THE FIRST OF THREE ARTICLES OUTLINING THE POTENTIAL SHORT TERM IMPACT ON AI ON SPECIAL NEEDS EDUCATION IN THE UK AND HOW THIS WILL BE TRANSLATED INTO SPECIFIC, CONCRETE, USEFUL TOOLS FOR TEACHERS THROUGH THE EARWIG SOFTWARE.
AT THE END YOU WILL BE ASKED TO LET US KNOW WHICH, OF A SELECTION OF FIFTEEN NEW MODULES AND UPGRADES, YOU THINK WILL BE THE MOST USEFUL.
The Squeeze
Most people outside education will be surprised to learn that, in fact, teachers spend most of their time not teaching.
According to the DfE's 2019 Teacher Workload Survey, teachers and middle leaders reported spending an average of 21.3 hours teaching out of their 49.5 self-reported working hours. That means that the act of teaching, on average, only took 43% of the working hours of the responding teachers. The rest of the teachers' time was spent on planning and preparation, marking and assessment, analysis and evaluation, pastoral duties and behaviour management, parental communication, classroom management, administrative tasks, professional development, extracurricular activities, and collaboration with colleagues and external agencies.
How much time and effort goes into these ancillary activities depends on many factors. But there is no doubt that, in the special needs environment, not only is there more time spent on the caring element, there is also considerably more of the other stuff. This is principally because every pupil has different needs and performs at a different pace. So learning and progress recording has to be as individual as each pupil.
In the UK, the individualisation of special needs education is crystalised in the Education, Health and Care Plan which is now how care for children with special needs is coordinated. This includes a set of desired educational outcomes agreed annually among all the interested parties, particularly the local authority required to provide the necessary resources. So special needs teachers have to report, in different ways to five different stakeholders
- School governors
- Trust leaders
- Parents
- The local authority
- Ofsted
Each of these stakeholders will examine performance from a different perspective and, therefore, require different sets of information.
So, the teaching element is constantly squeezed and the potential teacher work and stress levels increase with each squeeze.
What has technology brought to the party?
Since the arrival of the computers in the 80s, teachers have harnessed specific edtech and also general office technology to help them to reduce their workload and produce a more professional and instructive set of data for each of their stakeholders. However, it is still not unknown to find teachers printing photos of school activities, cutting them out and sticking them into exercise books.
As might be expected, the adoption of technology in schools is usually led by staff at mainstream secondary schools, with the processes filtering down to primary mainstream and then to special needs. But, because this individualisation of teaching and learning in the special needs environment places a heavier per pupil workload on SEN teachers, the opportunity for technology to provide productivity improvements is even greater in this sector than elsewhere in education.
Estimating the time saved, currently - in 2025 - by teachers, assistants and also the school leadership, using apps like this, is very difficult. But, anecdotally, a saving of about 20% to 40% for all levels of teaching staff, when compared with the manual methods being used fifteen years ago, would probably not be too far away.
This has helped schools to cope with both the expansion of per-pupil work in the post-Rochford era and the expansion in the number of pupils that they cater for, without similar-sized expansions in their budgets.
With an increasing proportion of pupils being diagnosed with special needs every year, this pressure will continue to increase, while budgets probably won’t.
What impact will AI have?
Over the last 20 years, the technology has evolved to carry an increasing share of the teacher’s admin workload - and Earwig has been at the forefront of this. But this has been using ‘conventional’ technology and the arrival of AI has transformed both the range of possible development options available and the degree of assistance that these can provide to educators at every level - to the point where it is possible to envisage AI providing each teacher with the equivalent of a hugely capable, knowledgeable and efficient assistant at their side, carrying out all the mundane tasks and even giving lessons.
That we will get there is not really in doubt. The question is - how do we get there, and is the ride going to be smooth or bumpy?
Many school leaders will have now had some personal experience of using AI powered tools - and the experience to date is not always positive - as anyone who has ever tried to deal with an AI powered customer service chatbot will attest!
My experience of many of the AI powered apps currently on offer goes like this…
‘Wow’ will it really do that? The demo looks pretty impressive. Maybe I should use this.
Clearly it’s clever but not perfect. So have to make allowances and correct as we go along.
Is this worth the effort? Maybe I could do it better myself?
Uninstall the app.
So, how do we avoid this in educational software - bearing in mind that we are working often with people who are generally not tech-oriented?
Let’s start with what AI already does well. It plays chess well (Intuitive). It writes text well, sometimes (Generative).
One of the other things that AI already does well is to look at a limited data set – e.g. the information about an individual pupil - and compare this with a much larger data set - e.g. the information about lots of pupils in a similar situation and come up with conclusions and suggestions.
We think that it could be this ability to learn about each individual and make specific suggestions, which could provide some really useful assistance to teachers and thus help turn the trick in special needs teaching and show real value to sceptical teachers, early on.
How is this likely to be implemented, in practice?
Currently, the teaching departments of special schools buy software to help their staff to handle some or all of these tasks…
- planning and resourcing lessons,
- assessing pupil performance,
- baselining prior attainment and tracking progress over time,
- setting end of year expectations or EHCP/IEP targets,
- evidencing teaching and learning,
- producing school reports,
- engaging and empowering parents,
- analysing performance trends for individuals and cohorts, monitoring behaviour and recording incidents,
- raising and managing safeguarding or wellbeing concerns, and
- mapping and tracking interventions and other provisions.
It is possible to envisage AI being harnessed to carry much of the load in every one of these activities within five years - and most of the load within ten.
What would be the financial impact?
Will this mean that teachers are no longer required? Of course not. The human interaction is vital to the development of children. But it should mean that, in special needs, individual teachers can deal with more pupils at any one time.
If the number of pupils that the average special needs teacher can deal with increases by 50%, this will mean a saving of more than £3billion annually - or, more likely, that schools will be better able to absorb the increase in SEN pupil numbers that will, almost certainly, happen over the next few years.