AI is becoming increasingly common across funded learning, but its value depends on more than the technology itself. To provide meaningful support, it needs access to accurate information, an understanding of the learner’s wider circumstances and clear safeguards around how it is used.
AI can produce fast, polished and confident answers.
That makes it tempting to assume that adding an AI assistant to a learning platform automatically gives it an understanding of the learner, their programme and the organisation supporting them.
But it does not – AI only knows what it has been allowed to access. The quality of its support depends largely on whether it can see enough of the wider picture.
Imagine you are asked to advise a learner and you’re given only a tiny amount of information: the specific question they’re asking.
“I’m stuck on this task. What should I do next?”
Without any additional context, you might offer encouragement, explain the topic or suggest breaking the task into smaller steps.
That advice is likely to be reasonable, but it’s also likely to be generic, and possibly overlook something important.
Now imagine you also know that:
The support you can provide to the learner now is likely to be far more useful.
Instead of giving some generic advice, you might say:
“This task is asking you to demonstrate how you prioritise competing pieces of work. You used a similar example in your last activity, but your tutor asked you to explain your decision-making more clearly. Choose one real workplace example, describe what needed to be done, explain how you decided what to prioritise and outline the result. This could also provide useful evidence for your progress review next week.”
As you can see here, the improvement doesn’t come from better wording, but from better information, and the same is absolutely true of AI assistants.
It’s important to remember just how convincing AI can be. It can sound helpful even when it is working with limited information.
Let’s consider another learner question:
“Am I on track?”
A generic AI tool might suggest checking deadlines, staying organised and speaking to a tutor.
But as we know, in funded learning, being “on track” means something quite specific. A reliable answer may depend on:
Without this information, an AI assistant can offer some reassurance or standard organisational and time management advice. It can’t, however, reliably assess the learner’s position.
With the right context though, it can provide a more meaningful answer:
“You are up to date with your planned learning activities, but two pieces of evidence are still awaiting submission. Your next progress review is in ten days, so completing them beforehand would keep you aligned with your programme plan. It may also be helpful to revisit the feedback from your previous review, where your tutor identified reflective writing as an area for development.”
Again, the difference is not simply the intelligence of the tool, but the quality and relevance of the information that is available to it.
Your staff need context too
The value of context extends way beyond learner support.
Think about a tutor preparing for a progress review. The information they need may be spread across multiple systems, including:
An AI assistant with access to only one part of that information may produce a well-written summary while still missing really important and relevant topics that need to be covered in the review.
But an assistant that can securely bring together those relevant records could provide a much stronger starting point:
“Since the previous review, the learner has completed three planned activities and submitted evidence for two competencies. Engagement has remained consistent, although one assignment is seven days overdue. Their employer has reported improved confidence when dealing with customers. Written reflection was identified as a development area at the previous review, but it has not yet been revisited.”
This does not replace the tutor’s professional judgement, but it reduces the time spent gathering information and helps the tutor to really focus the conversation on the learner’s progress, needs and next steps.
In many provider organisations, important information is stored across several systems.
Enrolment details sit in one platform, learning activity is captured in another. Reviews, evidence, employer communication and reporting may all be managed separately.
Staff often become skilled at piecing this information together, but AI cannot do that unless the technology has been designed to give it secure and appropriate access.
When an AI tool can see only one system, one document or one interaction, it sees only one part of the learner’s story.
And even where organisations do hold a large amount of learner information, it doesn’t necessarily mean that data is consistent or can be interpreted by AI in the most effective way.
Highly configured systems or connected, modular systems can store similar information in multiple different ways. The same type of data may be labelled differently, captured in different fields on different systems, or recorded in different formats across internal teams.
When AI mines data across a learner’s history in order to identify patterns or compare information over time, it will struggle to distinguish between genuine patterns if the underlying data is inconsistent. This can lead to missed connections or confident, yet inaccurate, conclusions.
Of course, assistive intelligent technology is still going to be useful for:
These are simple, yet valuable uses.
But more meaningful support does require a broader understanding of what is happening.
For example, identifying that a learner may need additional support is rarely a matter of noticing one missed task. It may require recognising that the learner has also missed a progress review, stopped engaging with learning activities and is approaching an important programme milestone.
Each of these signals may well appear insignificant on their own, but together, they may indicate that someone needs to take a closer look.
The solution is not to give AI access to everything, because useful AI also requires clear boundaries.
A helpful comparison to consider is a new colleague joining your organisation.
You would give them the information they need to perform their role. You would explain their responsibilities. Their account permissions would restrict any access to sensitive records. You would (initially, at least) review important work they completed before it was acted upon, and in the early days you would also expect them to show you where their information came from.
AI assistants, just like new colleagues, should be treated with the same care.
So when implementing AI assistants, it’s really important for organisations to consider:
And of course, the level of control should reflect the importance and risk of a task.
Using AI to improve the wording of a sentence is very different from using it to identify a compliance risk, influence a learner-related decision or recommend an intervention.
The greater the potential impact, the stronger the need for oversight.
AI assistants, chatbots and co-pilots are becoming much more common across funded learning. It’s important for us to be fully cognizant of what they understand. If you're considering introducing AI capability as a provider organisation, ask yourself:
These questions matter because effective AI is not simply something an organisation can purchase and switch on.
Its value depends on whether the organisation’s technology, data and working practices can support it.
AI capabilities in their various guises are becoming fairly standard across learning technology, and soon every platform will offer an assistant in some form or another.
So what will make a difference to providers is not really whether their learner management system offers an assistant, but whether that assistant can access the right information, understand the context in which it is operating and work within clear and appropriate boundaries.
When AI sees only a single question, it can provide a plausible response.
When it understands the learner, their programme, their progress and the requirements they are working towards, it can provide something far more useful.