Innovation

This project explores the use of AI-powered holographic conversational agents for teaching, outreach and public engagement. Using life-sized Proto displays, the team created interactive educational avatars based on figures such as Ada Lovelace and Averroes, allowing learners to engage in natural dialogue with historically significant figures.

The project combines educational design, source-grounded content and safety checks, with deliberate attention to inclusion, voice, accent and knowledge sources.

Why it matters

Holographic AI tutors can create a stronger sense of presence and offer new ways to support curiosity, dialogue and inclusive educational design.

This project examines how a life-sized holographic conversational agent can support group learning in a real university classroom. In a Professional Issues in Computing workshop, students interacted with a holographic Averroes avatar as part of a structured case-based discussion task.

Student feedback suggested that the system was easy to use, relevant to the activity and helpful in supporting group conversation.

Why it matters

This work shows how AI can move beyond one-to-one screen-based interaction and act as a shared learning partner in classroom discussion.

This project explores how AI-generated formative feedback can be integrated into programming quizzes in Moodle Learn using CodeRunner. The approach combines automated test cases with a custom CodeRunner template, allowing students to receive targeted guidance when their answers are incomplete or incorrect.

The system is used across weekly activities, lab exercises, mock tests and assessed work in Introduction to Programming and Databases modules. It supports formative learning by helping students review and improve their code more independently, without replacing the existing marking workflow.

This year, the system has generated 11,626 pieces of feedback, including around 2,000 AI-generated responses. The integration has also supported lab assistants by giving students quicker and more consistent feedback during programming activities.

Why it matters

Programming students often need immediate feedback while they are still working through a problem. This project shows how AI can be carefully embedded into existing learning platforms to provide scalable formative support, while keeping assessment processes and academic judgement in place.

This project explores how real-time learning analytics and smart glasses can support teaching, feedback and intervention in computer lab sessions. Developed within the Centre for AI in Education at Loughborough University, the framework brings together a lecturer-facing dashboard and a smart glasses interface for lab assistants.

The dashboard provides academics with an overview of where students may be struggling, which questions are causing difficulty and where support may be most useful. Lab assistants can then receive focused, anonymised support cues through smart glasses while moving around the lab and working with students. These cues can help assistants identify common issues, prioritise support and offer more targeted guidance without needing to repeatedly check a separate device.

The aim is to keep learning analytics visible to staff rather than students, while helping teaching teams provide timely, targeted and privacy-aware feedback during practical coding activities. The framework has been established and will be piloted in the next academic year.

Why it matters

Computer lab teaching can make it difficult for staff to see where support is needed in real time. This project shows how lecturer-facing analytics and smart glasses can work together to help staff respond more quickly and consistently, while keeping student support discreet, practical and focused on learning.

This project explores how generative AI can support educators in turning curricular content into engaging educational stories, while keeping human expertise central. It introduces an AI Storytelling Framework based on an eight-step process and a Right Human-in-the-Loop approach, where the right experts review content at key stages.

The framework was applied through the Named after Nelson podcast series, developed in collaboration with the Nelson Mandela Foundation, showing how AI-supported storytelling can produce educational resources that are creative, accurate and culturally responsible.

Why it matters

This work offers a practical model for using AI in educational content creation while keeping quality, accuracy and inclusion at the centre.

Using AI in cultural storytelling brings opportunities, but also risks around accuracy, representation, voice and emotional tone. This project develops a governance approach for AI-assisted storytelling that places expert human review throughout the content lifecycle, rather than treating it as a final check.

The work shows how clear review points, expert roles and documented decisions can support more trustworthy AI-assisted educational media.

Why it matters

Cultural storytelling needs care, context and accountability. This project helps institutions and educators use AI while protecting cultural integrity and reducing representational harm.

Providing timely and personalised feedback is one of the major challenges in higher education. This project explores how AI can support formative feedback through interactive quizzes in a Moodle-based learning platform, while keeping academic judgement in the hands of educators.

The system was tested across four university modules and generated more than 10,000 interaction records. The work also identified areas that need careful design, such as avoiding overly long responses and ensuring that students are supported without being directed too strongly towards model answers.

Why it matters

AI can help provide scalable formative support when it is carefully embedded into teaching practice and reviewed by educators.

This project develops a human evaluation rubric for assessing AI-generated language outputs. Rather than relying only on standard accuracy-based tests, the rubric considers linguistic quality, style, terminology, content accuracy, cultural appropriateness and logical coherence.

The work supports more inclusive and context-aware evaluation of AI systems, particularly for open-ended educational and professional writing tasks.

Why it matters

AI evaluation needs to account for language, culture, terminology and meaning, not only whether an answer is technically correct.

This project explores how large language models can support expert knowledge collection while keeping the process controlled, auditable and accountable. It introduces MHAESTRO, a hybrid two-stage approach that combines expert-designed decision trees with AI-supported conversational interaction.

The system allows AI to make the interaction feel more natural, while the structure and direction remain governed by expert input.

Why it matters

This work offers a practical way to combine conversational AI with expert-designed pathways, supporting more reliable use in education, evaluation and organisational decision-making.

Generative AI is changing what polished coursework can show about student learning. This project develops a practical framework for assessment design that places greater value on evidence of thinking, authorship, judgement and process.

The work considers assessment legitimacy, educational integrity, disclosure, moderation and programme-level governance, without relying only on detection or a return to exams.

Why it matters

Universities need assessment approaches that remain credible, fair and inclusive in an AI-rich environment. This work supports assessment design that focuses on evidence of learning rather than suspicion alone.