Aston University and Electronic Arts Limited KTP 24_25 R4

A game developer types a line of code, and an AI tool instantly suggests the next block—or flags a bug before it crashes the build. This project between Aston University and Electronic Arts builds bespoke generative-AI tools that automate parts of game software development, using machine learning and data mining to optimise how developers spend their time. The problem is straightforward: making video games is slow, repetitive work. Developers write vast amounts of boilerplate code, test the same scenarios repeatedly, and hunt for bugs that a machine could spot faster. Current tools are generic; this project aims to create AI that understands the specific structure and logic of a game’s codebase, then acts on it—generating code, suggesting fixes, or automating tests. If it succeeds, the impact is efficiency. Game studios could ship titles faster, with fewer crashes and glitches, while developers focus on creative design rather than drudgery. The tools could also transfer to other software industries—anywhere that writes complex, iterative code. The abstract does not give cost or timeline figures, so the scale of adoption remains unclear, but the principle is concrete: teach AI to read a game’s code, then let it write the boring parts.