Using AI to understand and improve eyewitness memory performance

My research proposes to better understand and improve human memory by enhancing police lineup procedures using Artificial Intelligence (AI) like Face Recognition Systems (FRS) and Large Language Models (LLMs). The goal is to create fairer lineups, reducing biases and misidentifications inherent in traditional methods.

By integrating psychological insights on memory and identification with AI's precision, I aim to improve the discriminability and reliability of eyewitness identifications.

This involves developing Python software, pyWitnessAI, to handle various tasks, including analysing videos and generating lineups with different strategies.

The research will explore several hypotheses on AI-generated lineups based on physical similarity, verbal descriptions, GPT-generated portraits, and comparative studies to understand the effectiveness of AI in lineup construction. Additionally, I will address ethical considerations and limitations of AI in this context.

My work seeks to set a new standard in eyewitness identification methodology, combining psychological principles with technical advancements to better understand and improve human memory.