Today we audit gauges, PFMEAs, Control Plans and special characteristics.
Tomorrow, we may also need to audit training data, AI models, drift, explainability and algorithm bias.
Recently, VDA published a draft yellow book on Artificial Intelligence in Quality Management with feedback open until June 30.
This is an important development because it signals that AI is no longer being seen only as a productivity tool.
It is now being treated as a quality-relevant system that must itself be controlled, validated, documented and audited.
The document highlights several new questions quality teams may soon need to answer:
• Who trained the AI model?
• What data was used?
• Was the data complete, representative and reliable?
• Can the AI decision be explained?
• How is model drift monitored over time?
• What happens if the AI gives a wrong classification?
• Who approves model re-training?
• How do we ensure traceability of AI-based decisions?
The yellow book connects AI with existing quality systems such as ISO 9001, IATF 16949 and VDA 6.x.
It specifically highlights areas such as:
• Auditability of AI models
• Explainability of decisions
• Bias in training data
• Hallucinations and incorrect outputs
• Predictive and prescriptive quality
• Anomaly detection
• Robustness and trustworthiness
• Integration of AI into supplier audits, process audits and product audits
One interesting point is that the document even links AI risk levels with functional safety, SOTIF, special characteristics and cybersecurity relevance.
This means the future quality engineer may need to understand both MSA and machine learning.
The future auditor may not only ask whether the gauge was calibrated.
They may also ask whether the model was trained correctly.
For those working in automotive quality, manufacturing, supplier quality, digital transformation and AI, this draft is worth reviewing before the June 30 feedback deadline.
Quality is entering a new phase where we will not only manage process variation.
We will also need to manage algorithm variation.
#AI #QualityManagement #IATF16949 #VDA #ArtificialIntelligence #ManufacturingQuality #AutomotiveQuality #DigitalTransformation #MachineLearning #QualityEngineering
