Imagine an AI acting like a biased judge who makes unfair decisions based on a person's background. This work creates a digital 'health check' and a repair kit to find those hidden prejudices in the software. It's like having a spell-checker, but for fairness, ensuring the AI treats everyone equally before it's used in the real world.
Companies using AI for decision-making face legal and ethical risks if their systems amplify social inequalities or discriminate against certain groups. There is a lack of concrete technical tools to detect, measure, and fix these biases before deployment.
A software suite featuring a bias diagnosis engine, a reparation and mitigation engine, and a synthetic data generator for testing AI fairness.
If you are a medical software provider dealing with biased patient triage algorithms — this project developed a diagnosis and reparation engine that identifies and removes unfair biases in healthcare data and algorithms.
If you are a recruitment platform dealing with discriminatory hiring filters — this project developed a Fair-by-Design methodology that ensures AI-driven candidate selection is impartial and rights-respecting.
If you are a public agency dealing with AI for vulnerable groups — this project developed a controlled experimentation environment that allows you to test and validate AI fairness on sensitive data on-premises.
Based on available project data, no specific pricing or cost models are mentioned; however, the tools are accessible via the AI-on-Demand platform and as an on-premises prototype.
Yes, the project includes an on-premises prototype for privacy-preserving tests and is being integrated into the Italian 'AI Factory' for pre-commercial validation.
Based on available project data, specific licensing terms are not provided, but the project delivers software releases of the final AEQUITAS framework.
The project provides a reference for regulatory sandboxes in Member States and offers a governance environment to ensure compliance with fundamental rights and fairness principles.
It is available as a service sub-component via the AI-on-Demand platform or as a stand-alone release for on-premise installation to protect sensitive data.
The consortium is well-balanced for commercial transition, featuring 19 partners across 8 countries. With a 32% industry ratio (6 industrial partners), the project blends academic research from 5 universities with practical application, ensuring the tools are not just theoretical but tested against real-world business constraints.
Contact ALMA MATER STUDIORUM - UNIVERSITA DI BOLOGNA
Contact us to explore how to integrate the AEQUITAS fairness toolkit into your AI pipeline.
