Mr Wei Shen1
1Griffith University, Southport, Australia
Biography:
Wei Shen is an eResearch Analyst at Griffith University, where Wei helps researchers adopt digital tools and technologies to enable and enhance their research. With over 17 years of experience in the higher education sector and a world-class private research lab, Wei brings a systems-thinking approach to problem-solving, grounded in deep knowledge of research infrastructure and digital solutions.
Wei is particularly interested in helping researchers navigate the hype surrounding artificial intelligence—making it approachable, safe, and genuinely useful, especially in academic environments where context, accuracy, and trust are critical.
Abstract:
This presentation shares an honest journey of trial, error, and learning while trying to build an AI-assisted policy finder for researchers. What started as a simple idea—using AI to help users locate relevant, more importantly, complete government policies—quickly became a series of unexpected challenges.
Wei experimented with a wide range of AI tools and techniques, from no-code solutions to standalone large language models (LLMs) and retrieval-augmented generation (RAG). Each method promised to simplify the task, yet all struggled with real-world issues like context loss, inconsistent responses, and difficulty aligning outputs with user expectations. In particular, Wei will reflect on the limitations of popular tools and why even technically “correct” implementations can still fail to meet practical needs.
Rather than showcasing a polished solution, this talk offers practical lessons learned from what didn’t work – valuable for anyone building AI tools for institutional knowledge, policy search, or similar domains. Attendees will gain a realistic view of the current capabilities and pitfalls of AI-powered search, and hopefully, avoid some of the detours I took along the way.