Artificial intelligence is everywhere, but how do you separate the hype from the helpful when it comes to risk models?

The “AI in Quantitative Risk Measurement” track at SIRAcon ‘25 is all about the practical side of integrating AI and machine learning into your risk analysis. Not to build flashy toys—but to make your models faster, smarter, and more insightful.

This track isn’t about automating your entire job away. It’s about enhancing your ability to detect patterns, handle uncertainty, and generate defensible, data-driven insights with greater efficiency. If you’ve ever wondered whether a model could learn from past scenarios, detect bias in your inputs, or help you forecast risk with fewer assumptions, we think you’ll find some answers here.

AI that Adds, Not Obscures

In risk, clarity matters more than complexity. That’s why these sessions will focus on interpretable AItools and techniques that enhance your understanding instead of burying it in black-box algorithms.

Potential session topics include:

• Leveraging machine learning to spot patterns in incident and loss data

• Using natural language processing (NLP) to extract risks and controls from unstructured sources

• Building AI-enhanced models that provide explainable outputs—not just predictions

• Identifying when AI helps… and when it just adds noise

Because if you can’t explain how your model works, you probably can’t defend it either.

Faster, Leaner Modeling with ML

AI can do more than just analyze data. It can accelerate how you prep, structure, and test your models. We’re hoping to hear from experts on time-saving applications of machine learning that let you skip the grunt work and get to insight faster.

Learn how to:

• Auto-clean and cluster messy data sets from multiple sources

• Use AI to flag anomalies and inconsistent assumptions before they skew your results

• Train models on historical loss data to inform probability distributions and impact ranges

• Optimize Monte Carlo simulations with smarter parameter tuning and convergence detection

No data science background required—just a low tolerance for cleaning up messy models.

Redefining Risk Scenarios with AI

Scenario development is the heart of good risk quantification, but it’s also one of the most labor-intensive steps. We’d like to hear from experts on how AI can augment scenario creation with smarter suggestions, better inputs, and real-time threat intelligence integration.

Expect to see talks covering:

• Tools that generate realistic risk scenarios based on current threat trends

• Ways to blend AI-generated insights with SME judgment (without losing control)

• Techniques for continuously updating scenarios as new data becomes available

• Risk storylines driven by both structured and unstructured sources

The result? More relevant scenarios. Less mental gymnastics.

AI + Human Judgment = Better Risk Decisions

AI isn’t here to replace you. As long as you approach it properly, it’s here to supercharge you. The real value of AI in quant isn’t in outsourcing thinking, but in enhancing it. Sessions in this track will help you strike the right balance between automation and expert oversight, and show you how to turn AI outputs into decisions leadership can trust.

You’ll walk away with:

• A working knowledge of AI’s strengths and limits in risk modeling

• Practical tools to experiment with right away (no PhD required)

• New ideas for solving old problems like data gaps, stale assumptions, and overworked analysts

Build Better Models, Make Smarter Moves

The “AI in Quantitative Risk Measurement” track is designed for risk professionals who want to move beyond spreadsheets and truly level up their modeling practice. Whether you're just starting to explore AI or already experimenting with it in your workflows, this track offers real-world insights, not just research papers.

At SIRAcon ‘25, we’re not just talking about the future—we’re putting it to work.