Challenge 1: From "Huh?" to Aha! Cracking the code of research papers

Subject Area: Artificial Intelligence

Supervisory Team: Audrey Ekuban and Venetia Brown

Scholar: Oke Onyebuchukwu

How can we make studying more engaging and interactive?

Meet Oke, whose research transforms research papers into interactive learning experiences.

Oke came to KMi during the summer while studying Mathematics and Computer Science at university. His project focused on ways to transform student engagement with research papers by developing the Research Companion app. Oke used Retreval-Augmented Generation (RAG) technology and prompt engineering to simplify access to academic literature by using functionalities such as summarisation, visual aids generation, quizzes and a chatbot. His evaluation of the chatbot's performance across different temperature settings reveals that higher temperatures yield more consistent and reliable outputs. However, the quality of responses is dependent on the retrieval process.

Empowering students with generative AI to support academic reading. Oke's research is important in providing more inclusive educational tools that leverage AI to support the diverse learning needs of students. Using the Research Companion App makes research more interactive and accessible and might help those struggling with reading activities engage more easily with complex materials.

Check out Oke talking about his experiences on the scholarship here, and watch the demo of his app here. His report can be read here.

Concept Overview

Interview

Oke designed an app to assist students in understanding research papers through interactive features. He adapted very well to what was new content for him. He proved to be analytical, suggesting new approaches to improve the chatbot's performance and robustness.
- Audrey Ekuban, Supervisor

Don't worry about not having advanced programming experience. The team provides a lot of helpful resources to get you started. You can have a flexible work schedule. I would recommend it.
- Oke Onyebuchukwu, 19 years old. 2024 award winner and scholar

Challenge 2: Building Trust with AI: Can CORE-GPT Be Our Truthful Friend?

Subject Area: Open Access to Research Papers

Supervisory Team: Matteo Cancellieri, Suchetha Nambanoor-Kunnath and David Pride

Scholar: Ian Obiogbolu

Will academic research be revolutionised by AI by improving access to factual answers and open access to knowledge?

Meet Ian, who explores how CORE-GPT and different LLMs handle research questions in various disciplines.

Ian studied on an Access to HE course before joining the summer scholarship. His project investigated how various Large Language Models (LLMs) interact with CORE-GPT, a specialised question-answering platform that provides answers from credible scientific literature. Ian used three metrics to evaluate different models, testing their comprehensiveness, trust and utility and comparing the responses.

Enhancing accessibility and impactful research. Ian's project empowers research students to rely on accurate and trustworthy information for their studies and projects while establishing the most compatible and cost-effective solution for future CORE-GPT applications. Check out Ian demonstrating the platform here and his interview, talking about his experience on the scholarship here. His report can be read here.

Concept Overview

Interview

Ian did his scholarship with the CORE team, and he helped us look into how various large language models (LLMs) interact with CORE-GPT. Working with LLMs poses challenges in terms of the accuracy of the answers, costs, and load on the system. Ian's work helped us understand what are the most promising models we can use to integrate the new generation of CORE-GPT.

He proved to be dedicated and with great attention to detail; it was a pleasure working with him, and we are planning to use his report for a CORE blog post and kick-start future work on our new version of CORE-GPT.
- Matteo Cancellieri, Project Supervisor

Applying for the scholarship gives you a chance to work with experienced developers and researchers, work with real-world technologies and impact the world of education.
- Ian, 19 years old, 2024 award winner and scholar.

Challenge 3: Generative AI and Controversial Topics

Subject Area: Artificial Intelligence

Supervisory Team: Aisling Third, Tracie Farrell and Matteo Cancellieri

Scholar: Prince Akah

How do AIs deal with disagreement?

Meet Akah, who generated AI "participants" with polarising viewpoints on controversial topics and assessed the quality of arguments.

Prince joined the scholarship after studying on an Access to HE (Computing) course. His research assessed how different large language models (LLMs) generate content about controversial topics. Those who have engaged with LLMs know that producing a definitive point of view about polarising topics is challenging. Prince was interested in studying the realism and quality of the arguments LLMs present when they are tasked with defending a particular point of view. Prince attempted to generate a selection of debates in different formats, including two AI "participants" designed to have polarising viewpoints on controversial themes. He experimented with different prompts around the personas given to the LLMs, including various details about their supposed experience or identities, to see how the LLM would bring this experience to the debate and how this impacted human judgements about realism and quality.

Navigating controversy and promoting critical thinking. Prince's project helps to understand and identify features humans use to judge whether or not a piece of content has been generated by AI, the quality of AI-generated arguments, and the extent to which those arguments are convincing. By exploring how AI models handle disagreements, students could learn to critically evaluate different perspectives and develop informed opinions.

Check out Prince's interview about his experience on the scholarship and a demo of his work below.

Concept Overview

Interview

Working with Prince was a pleasure. He chose a very interesting area of research that we suspect will only increase in relevancy in the coming years and showed that he can spot a good juicy problem. We also found him to be creative and curious, which are all excellent qualities in a researcher, and we hope that he continues to develop his skills in computing.
- Tracie Farrell, Project Supervisor

This scholarship offers an incredible opportunity to work on cutting-edge projects and gain hands-on experience with AI technologies. It's a chance to challenge yourself, explore new ideas, and contribute to meaningful research. Don't hesitate to apply if you're passionate about technology and want to make a real impact.
- Prince Akah, 19 years old, 2024 award winner and scholar.