Oberon Buckingham-West
Academic and research departments
Surrey Institute for People-Centred Artificial Intelligence (PAI).About
My research project
Anti-Atrophy Agents: Preventing Skill Loss in an AI-Automated WorldOberon Buckingham-West is a PhD Researcher at the Institute for People-Centred AI (CVSSP), specialising in Multimodal Agents and Human-AI Interaction.
His research investigates interaction paradigms that prevent skill atrophy in AI-assisted creative workflows. Specifically, research focuses toward fine-tuning Vision-Language Models (VLMs) and Diffusion models to create process-based personal tutorials that support, rather than replace, human creativity.
Leveraging over 10 years of industry experience in Audio Engineering and Game Development, Oberon brings a practical, accessibility-first perspective to tool design. He is currently exploring ground-truth data pipelines to benchmark model performance on noisy UI data. Beyond his research, he also contributes as an AI Education Product Tester for DeepLearning.AI and frequently speaks on video game panels, such as guildford.games and Game Dev Local.
Supervisors
Oberon Buckingham-West is a PhD Researcher at the Institute for People-Centred AI (CVSSP), specialising in Multimodal Agents and Human-AI Interaction.
His research investigates interaction paradigms that prevent skill atrophy in AI-assisted creative workflows. Specifically, research focuses toward fine-tuning Vision-Language Models (VLMs) and Diffusion models to create process-based personal tutorials that support, rather than replace, human creativity.
Leveraging over 10 years of industry experience in Audio Engineering and Game Development, Oberon brings a practical, accessibility-first perspective to tool design. He is currently exploring ground-truth data pipelines to benchmark model performance on noisy UI data. Beyond his research, he also contributes as an AI Education Product Tester for DeepLearning.AI and frequently speaks on video game panels, such as guildford.games and Game Dev Local.
University roles and responsibilities
- PhD Researcher part of the first cohort in the CDT in AI for Digital Media Inclusion.
My qualifications
Affiliations and memberships
News
In the media
AI CO-LEARNER FOR EDUCATION | Bridewell Royal Hospital Foundation
AI CO-LEARNER FOR EDUCATION | Bridewell Royal Hospital Foundation
Collaborative Industry-challenge as part of the University of Surrey’s Centre for Doctoral Training in AI for Digital Media Inclusion (CDT), delivered in partnership with Royal Holloway, University of London. The project began in March 2025, with a team of AI researchers working alongside the school trust to help ensure that pupils can access and navigate the "Bridewell AI" initiative and roadmap with clear purpose.
ResearchResearch interests
My research sits at the intersection of Human-AI Interaction (HAI) and Multimodal Agents, specifically fine-tuning Vision-Language Models (VLMs) and Diffusion models to support creative workflows. I investigate interaction paradigms that prevent skill atrophy in AI-assisted environments by developing process-based personal tutorials rather than automated solutions. Additionally, I focus on building ground-truth data pipelines to benchmark model performance on noisy UI interaction data, ensuring that sophisticated tools remain accessible and pedagogically sound for diverse users.
Research projects
Predicting Game Engine Interface Changes from User IntentAn positional method to help users navigate complex graphical user interfaces (GUIs) in creative software like Unreal, Unity, and Godot. While existing Vision-Language Models (VLMs) and GUI agents often automate tasks (potentially hindering user skill development) this research aims to reduce cognitive load by guiding users. The proposed system takes a screenshot and a text description of user activity to predict the likely next screen region where a GUI change will occur.
Research interests
My research sits at the intersection of Human-AI Interaction (HAI) and Multimodal Agents, specifically fine-tuning Vision-Language Models (VLMs) and Diffusion models to support creative workflows. I investigate interaction paradigms that prevent skill atrophy in AI-assisted environments by developing process-based personal tutorials rather than automated solutions. Additionally, I focus on building ground-truth data pipelines to benchmark model performance on noisy UI interaction data, ensuring that sophisticated tools remain accessible and pedagogically sound for diverse users.
Research projects
An positional method to help users navigate complex graphical user interfaces (GUIs) in creative software like Unreal, Unity, and Godot. While existing Vision-Language Models (VLMs) and GUI agents often automate tasks (potentially hindering user skill development) this research aims to reduce cognitive load by guiding users. The proposed system takes a screenshot and a text description of user activity to predict the likely next screen region where a GUI change will occur.