The AI-RFX Procurement Framework

The AI-RFX is a procurement framework is a set of templates to empower industry practitioners to raise the bar for AI safety, quality and performance.

The framework is open source, and converts the the Principles for Responsible Machine Learning into a checklist.

Raising the bar for AI safety, quality and performance in industry

The AI-RFX procurement framework has been put together by a group of domain experts. Its purpose is to ensure best practices in industry during the procurement, design, devleopment and deployment of machine learning systems in industry.

This procurement framework goes beyond the machine learning algorithms. It provides a method to assess the maturity level of the technical infrastructure and processes around the algorithms themselves, by using our Machine Learning Maturity Model.


The Machine Learning Maturity Model

The Machine Learning Maturity Model is a set of criteria which ensures that the core technical infrastructure and the right processes are in place. It provides the ability for key industry practitioners to set a bar for quality, safety and performance when procuring machine learning solutions. You can request to download a copy of the AI-RFX Framework through the contact form below, which contains this document as well.

The model was designed using the Responsible Machine Learning Principles, and consists of the following 8 assessment criteria:

Number Assessment Criteria Responsible ML Principle
#1 Practical benchmarks Principle #6: Practical accuracy
#2 Explainability by justification Principle #3: Explainability by justification
#3 Infrastructure for reproducible operations Principle #4: Reproducible operations
#4 Data and model assessment proceses Principle #2: Bias Evaluation
#5 Privacy enforcing capabilities Principle #7: Trust by privacy
#6 Operational process design Principle #1: Human Augmentation
#7 Change management capabilities Principle #5: Displacement strategy
#8 Security risk mitigations Principle #8: Security risks

Contact us or join the Ethical ML Network (BETA)