Retaining personalisation with AI: How machine Learning preserves personalisation and the human touch

Personalisation and focusing on the individual child or young person’s voice within SEND (Special Educational Needs and Disabilities) services is crucial. An individualised person-centered approach ensures that children and young people and their parents/carers receive support tailored to their unique needs, granting them choice and autonomy over their lives.

As local authorities increasingly adopt digital and AI solutions, naturally concerns arise about the potential reduction in individualised person-centered care.

In this blog, we share our philosophy that AI and machine learning, combined with human-in-the-loop strategies, can enhance personalisation and improve outcomes for children and young people with SEND. Contrary to common fears, we believe these advanced technologies can complement and elevate individualised support within SEND services.

Let’s start by explaining personalisation with something we all love, coffee

Every day, rain or shine, Lucy visits her local coffee shop. The staff know her so well that they recognise her and remember her order. From autumn to spring, she gets a Black Americano, and in the summer, she switches to an Iced Flat White with Soya Milk.

This personalised service is a result of human interaction and customer insights gathered over Lucy's numerous visits. The baristas know her as a regular, remember her preferences, and adapt her order based on the weather.

However, this level of personalised service is difficult to scale. New staff members need time to learn individual preferences, and larger coffee shops with more customers and staff take longer to provide the same level of personalisation. It's also challenging for coffee shops to predict how individual variations in orders based on weather might interact with other customers' personalised orders.

For example, if Lucy is the only one who prefers a cold drink in the summer, the coffee shop doesn't need to change anything. But if everyone wants a cold drink in the summer, the coffee shop may need to adjust the ingredients they buy to avoid disappointing customers.

So, how do we scale this?

How Machine Learning is a fancy way of scaling personalisation

Machine learning models draw on large amounts of data. This data can be stored in central databases e.g. the way coffee shops use loyalty cards. Machine learning algorithms excel at rapidly analysing this data to uncover insights, which might otherwise take humans significantly longer to identify. This means that if different baristas arrive at the coffee shop, or in much larger stores, more customers can experience greater levels of personalisation.

However, the data used to drive this personalisation, still doesn't necessarily include insights that Lucy's favourite barista might know, like whether she wants a straw, or the precise temperature Lucy likes. This is why we believe a combination of AI and the human touch is important to personalise services.

Machine learning models are different to traditional models in their ability to quickly analyse vast amounts of data from multiple sources, making it quicker and easier to identify specific coffee preferences for customers. In the case of local authority SEND services, machine learning models can make it quicker and easier to identify children and young people who require an EHCP assessment, drawing on data from a broad range of sources, including written assessments using Large Language Models.

Machine learning models are also specifically able to identify patterns that are difficult to spot with traditional analysis. This capability helps in identifying specific interventions that are most likely to support a child or young person’s unique needs, enabling the right support to be provided earlier in their educational journey.

Predictive machine learning, which involves forecasting the future state, can also help services in making it more likely that they can personalise their offering. For instance, if we loop back to our coffee reference, this could involve forecasting a higher overall demand for iced drinks during periods of hot weather, ensuring the coffee shop has access to enough ice to meet customer demand. In the world of local authority SEND services, this could entail observing growing trends in the types of SEND and enabling local authorities or schools to develop sufficient skills or resources to support.

Let’s get to the nitty gritty around the importance of human-in-the-loop strategies for AI-systems

While machine learning can streamline personalisation and facilitate sharing insights with multiple staff members to meet large numbers of customer needs, it is important to maintain the human touch integrated into AI-driven solutions.

At Invision360, we believe in the following principles that help us integrate human-in-the-loop strategies into our AI-driven products:

1. AI should enhance, not replace, human interaction

While AI systems can process vast amounts of data, human insights are crucial for interpreting information and making empathetic decisions. Involving humans ensures that AI-driven recommendations align with individual values and needs. This is particularly important in the world of SEND services where children and young people may have specific preferences that are difficult to capture in data (e.g. preferred teachers or communication methods).

2. AI and ML systems should be designed to guarantee transparency, fairness and the ability to control and provide feedback on the system over time

Human-in-the-loop strategies involve a continuous feedback loop where human feedback refines and improves AI models. This iterative process ensures that AI adapts to changing needs and contexts. This is particularly important in the world of SEND services, where national policy changes can impact the types of provision available, or a children and young people’s needs may change over a number of years. Enabling professionals to provide feedback on AI-generated EHCPs ensures that AI remains a reliable support tool rather than an inflexible system.

3. AI-driven personalisation should feel authentic and considerate

Despite technological advancements, empathy remains a uniquely human trait. Multi-agency teams across SEND services can, and should, intervene at critical touchpoints to provide compassionate support and build deeper connections with children and young people and their parents/carers. For example, the way caseworkers share a draft EHCP with children and young people and their parents/carers may require a tailored and empathetic approach to comprehend and adapt to emotional, social, or other needs.

The Impact of AI on Personalisation

As the number of children and young people with SEND continues to increase, integrating AI and human-in-the-loop strategies offers a meaningful way to ensure we continue to deliver personalisation in the EHCP process. As machine learning algorithms become more sophisticated, their ability to support service providers to deliver highly personalised experiences will continue to improve. However, understanding and responding to individual human needs will always require a human element.

We believe that SEND services that successfully blend AI-driven approaches with empathetic, human-in-the-loop products will create personalised experiences that foster deeper connections with children and young people and their parents/carers. This can lead to improved outcomes and satisfaction in an increasingly digital world.

More articles you may like:

Privacy by design: ensuring safety with secure internal AI models for LLMs

Harnessing generative AI for EHCPs: A deep dive into Large Language Models

Improving EHCP Quality: The role of generative AI in supporting EHCPs