Episode two: The Power of scenario planning with ABMs and LLMs

The second episode of my AI generated podcast is focusing on the webinar I hosted for MentorCruise, “The Power of Scenario Planning for Businesses & Government”, recording is available on LinkedIn.

In this episode, the (AI-generated) hosts explore how businesses and governments can look beyond traditional predictive modelling by embracing the power of scenario planning. They highlight how typical machine learning relies heavily on historical data, causing it to fail when faced with unpredictable “black swan” events like the COVID-19 pandemic. To navigate these unknowns, the hosts introduce Agent-Based Modelling (ABM)—a framework that simulates interactions between individual agents to explore hundreds of hypothetical futures.

The podcast breaks down the fundamental building blocks of ABMs: the agents (which can be anything from people to households or even countries), the environment they inhabit, and the rules of behaviour that govern their interactions. By looking at famous examples like the Imperial College COVID-19 model and Schelling’s model of segregation, the episode illustrates how ABMs provide a digital “sandbox” to test policies and strategies that would be too costly or risky to trial in the real world.

Finally, the hosts delve into the modern evolution of the field: using Large Language Models (LLMs) as agents. This advancement allows simulations to emulate complex human reasoning, decision-making, and social dynamics far more realistically than ever before. However, while the potential is vast, the hosts conclude with a crucial caveat—the future of LLM-driven simulations depends entirely on grounding AI behaviours in empirical data to mitigate hallucinations and ensure the models remain trustworthy and representative