Synthetic Data Generation using
Augmented Language Models (ALMs)

Forum of Ideas | Dr Steve Bickley

Date:  Wednesday 15 May  

Workshop:  10am-12pm

Location: QUT Gardens Point

Event Details

Speaker Details

Steve bickley

For Part 1 of Steve's segment of the Forum of Ideas, we delve into the transformative journey of Augmented Language Models (ALMs) like OpenAI’s ChatGPT and the more recent Assistants API, and their expanding role in research analytics. Foundational technologies enabling ALMs include pre-trained models such as GPT-3.5-turbo and GPT-4, along with notable contributions from other providers like Google's Gemini and Anthropic's Claude series. These advancements have significantly elevated empirical study designs by providing augmented decision-making tools, particularly in synthetic data generation and silicon sampling. This session includes a hands-on exercise using SurveyLM, where participants in groups of 2-3 will simulate and analyze complex agent behaviors. This mirrors real-world decision-making using research methods and tools from behavioral economics, providing invaluable insights into the capabilities and applications of ALMs in modern research.

Steve is a trained Behavioural Economist and Research Fellow at the ARC Training Centre for Behavioural Insights for Technology Adoption (BITA Centre) at QUT, dedicated to understanding how real-world factors shape human behaviour and decision-making. His unique blend of professional backgrounds in electrical engineering, project management, and behavioural economics has shaped his distinctive approach to problem-solving. Steve’s research delves into the complex systems literature and beyond, offering fresh perspectives and innovative approaches to behavioural economics and social AI, aiming to enrich these fields by making social science research more robust and effective. His innovative research culminated in his PhD thesis, "Bridging Complexity and Behavioural Economics: The Constrained Methods Matrix," which presents novel tools and insights for modeling and analyzing dynamic and complex societal systems, especially during unforeseen events like disease outbreaks and technological transformations.