Iain Emsley

Mobile phone applications are increasingly embedding machine learning and generative Artificial Intelligence (AI) in daily life, such as identifying an object or wayfinding in an unfamiliar location. Frictionless installation hides both the layers of services needed for these activities to take place and the networks of companies and data flows that enable the software to exist. Software is a key medium to be researched, yet it is under-served in existing digital research infrastructure.
This project asks:
• how can we develop existing software to trace these services at scale?
• what are the challenges – such as environmental, ethical, and inclusivity – that require accounting for in these pipelines?
• what methods can be used to support the research communities and broaden their access to DRI?
This project traces the presence and histories of a set of AI services by reading Android application (APK) files. APK files are compressed files of machine code that are uncompressed on installation. As such, we need to use computational tools to engage with these files. Existing App Studies code will be scaled to run on HPC using disassembly tools, such as Androguard (a Python security engineering tool), to open them and enable researchers to look at links from malware packages and services, identify points of integration, and develop training material and documentation. This will be used to run workshops using this material with the App, Software, and Critical Code Studies communities to broaden access to and use of DRI for their research.