How AI allows us to support our learners better

By Sharmen Ibrahim, Group Director of Digital Education and Artificial Intelligence at Activate Learning

Artificial Intelligence (AI) is revolutionising the way we learn, access information, teach, and the landscape of educational assessment in general. This presents both exciting opportunities and significant challenges. AI offers the potential to streamline administrative tasks, enhance personalised learning, and generate contextualised feedback

However, the integration of AI in assessment comes with challenges. Ensuring the ethical use of AI, maintaining data privacy, and addressing biases in AI algorithms are critical concerns. Educators must be equipped with the knowledge and skills to effectively implement and utilise AI tools in their teaching practices.

It is crucial to remember that AI tools shouldn't replace human authorship and critical thinking skills. At Activate Learning, we ensure that our AI tools are designed with our learning philosophy at their core. We proudly refer to it as our human-centric approach.

The Activate AI Tutors we use personalise the learning experiences to meet learners' needs and help in reinforcing neural connections. By adapting to individual learning styles and needs, they ensure that students receive the right level of challenge and support to strengthen their cognitive abilities.

Moreover, as the Activate AI Tutors can provide real-time feedback, students see the direct impact of their efforts. This immediate reinforcement can boost motivation by showing students how their studies are relevant to their goals.

The tutors provide learners with non-judgmental feedback in real-time and by fostering a healthy atmosphere, they contribute to a positive emotional environment that is conducive to learning.

By combining the learning philosophy with AI, Activate Learning creates a comprehensive and compassionate learning experience. This integration ensures that students are not only academically prepared but also emotionally supported, ultimately leading to better educational outcomes and personal growth.

So how can we use AI while also safeguarding against malpractice? The key here is to focus on critical thinking. Educators can incorporate activities that require students to critically think, evaluate, and respond to questions on specific topics. They can also encourage learners to critically evaluate AI-generated content and use it as a starting point for further research and analysis. It is all about shifting the focus from the product to the process.

Educators also need to consider diversifying the way learners are assessed. Presentations, vivas, and projects are excellent ways to measure the learner's understanding of a topic.

As part of the design process, we evaluate, mitigate, and eliminate biases in AI by providing it with data that is diverse, representing various users’ backgrounds, abilities, and experiences. We also ensure that security measures are built-in to safeguard against potential threats and vulnerabilities from the outset.

So does it actually work? Yes, and we have the proof! Using variance inflation factor (VIF) and multiple regression analysis models, we looked at how well the usage of the AI Tutor predicts final grades and improvements in grades. We applied this to a group of 150 learners resitting their GCSE Maths exam in November 2024.

To ensure that the independent variables (the variables we think might affect the outcome) are not too closely related to each other, we used VIF. We then used multiple regression analysis to ensure that we accounted for other factors that might also affect grades, such as:

• Prior ability: How capable the student was before using the AI Tutor.

• Attendance: How often the student attended classes.

• Quiz engagement: How involved the student was in online quizzes.

What we found:

• The AI Tutor has a strong, positive effect on learner outcomes, with 10 uses being worth 1/3 of a grade level (then 100, then 1000).

• The AI Tutor is equally likely to be used by those with high and low initial assessment skills, and more likely to be used by those with exam access arrangements.

• Tutors have a strong impact on learner adoption of AI tutor tools.

We will continue developing the Activate AI tutors to ensure that improvements are made to the model and enhance the learner’s experience. We aim to run another analysis on a much larger data set once learners sit GCSE exams in June 2025.