L2M Validation / Qc Automne 2025 / AsbabAI is a novel platform that evaluates the causality of deep learning models to improve their explainability and robustness using causal inference,a key requirement in regulated sectors such as healthcare and finance

Artificial intelligence (AI) is increasingly used to make high-stakes decisions in fields like finance, insurance, and healthcare. However, most of AI systems function as “black boxes,” producing results without clear explanations. This lack of transparency can lead to serious consequences, such as denying loans to creditworthy individuals or making unfair risk assessments.

The AsbabAI aims to make AI models more transparent, reliable, and robust. Using advanced techniques called causal inference, it helps uncover the real reasons behind a model’s decisions and ensures those decisions remain valid even when real-world conditions change.

This project will be especially valuable for financial institutions, which face growing regulatory pressure to ensure fairness and explainability in their AI systems. It also supports public trust in AI and contributes to Canada’s leadership in responsible AI innovation.

Faculty Supervisor:

Sébastien Roy

Student:

Partner:

V1 Studio

Discipline:

Computer science

Sector:

Finance and Insurance; Artificial Intelligence

University:

Université de Montréal

Program: