Within the manufacturing enterprise there is no more paper-intensive area than the quality function. The very nature of inspection and verification requires the evaluation of paperwork and the documentation of observed results. This proliferation of quality paperwork spans nearly every aspect of the manufacturing process.

• Quality planning and risk avoidance through processes like APQP or FMEA generate volumes of data during product planning and introduction.
• Quality inspection of incoming material in preparation for manufacturing.
• On-line quality evaluation, control processes, planned inspections and response to event-based evaluation generate critical production-traceability data and the basis for continuous improvement.
• Post-production final inspection and non-conformance tracking from customer feedback add yet another layer of data and tracking requirements.

In many cases the legacy approach to this diversity was paper-intensive. Cross-functional requirements often resulted in point solutions for each organizational aspect. The end result was often a frequent yet unpredictable stream of human-errors in manual data collection, disjointed set of processes, work-around spreadsheet approaches, and poorly aligned data sets. In some cases individual quality functions have been automated but even then the data can be trapped in PLMs, MESs, Supplier Management and other purpose-built solutions.

The diversification of non-digital nature is prone to errors, time-consuming to manage, and hinders real-time insights. As manufacturing quality demands increasingly complex oversight, artificial intelligence (AI) could finally provide the catalyst for a more coordinated and consolidated approach across these systems. This shift promises to improve decision-making, reduce errors, and enhance product quality.

Opportunities to Transform Quality Management

AI - ranging from generative, predictive, and “simple” machine learning - is poised to address these challenges by automating data analysis, integrating multiple data sources, and providing real-time insights. Here’s how it can be applied in manufacturing quality management:

1. Automated data cleansing: Generative AI can automatically detect and fix common data quality issues like typos, missing values, and inconsistent formatting, significantly reducing manual effort required for data cleaning. By identifying and correcting errors in real-time, generative AI can help maintain high data quality across the entire data integration process.
2. Data mapping and integration: By analyzing patterns and relationships across different data sources, AI can map data fields and integrate data from various silos, creating a unified dataset for analysis.
3. Natural language processing (NLP): AI with NLP capabilities can interpret and standardize unstructured data like text, making it easier to integrate with structured data from other sources. Generative AI capability can be further extended to include non-textual data like photos and other images.
4. On-line Visual Inspection: Visual inspection may be the most prevalent use of AI in manufacturing. AI’s speed in data processing and ability to learn has already resulted in numerous applications detecting possible defects on surfaces or misaligned labels on high speed consumer product lines.
5. User Process Assistance: True quality starts with avoidance of mistakes and bad data. The development of bots to assist in quality planning / configuration, collecting inspection data and the execution of quality processes preventively attack the cost of quality.

The Arrival of the Right Tools for the Job

To be fair, these application opportunities are well understood challenges. There have been plenty of efforts that attack these problems with the automation of manual processes or complex master data management integration. Unfortunately, these approaches have proven difficult in terms of justifying the needed investment and the creation of adaptable solutions that could be vital in the long term. AI technology may prove to be the tool that manufacturing has been waiting for to deliver these benefits.

1. Breaking Down Data Silos: Breaking down silos gets much of the hype but it itself is not the goal. The goal is a unified view of the entire manufacturing process from planning through product delivery. An integrated approach ensures that all relevant data is considered, offering more accurate insights into quality performance.
2. Preemptive Improvements for Quality: Predictive insights have long been the holy grail of manufacturing whether it be in terms of avoiding equipment issues, variable deviation or product defects. Addressing quality issues before they occur or escalate is only possible with access to a complete and clean data set.
3. Better Resource Utilization via Automated Reporting: Much of the manual nature of existing quality systems centers around the documentation. This documentation may be required for compliance or genealogy tracking but it is an imperative. Generative AI can automate the generation of quality reports, removing administrative workload and the need for time-consuming manual entries.
4. Improved Decision-Making: AI can analyze large datasets to uncover insights that might not be immediately obvious to human analysts. These insights provide actionable intelligence that helps manufacturers make better decisions, prioritize corrective actions, and optimize production processes.

Conclusion

AI offers a significant opportunity to transform manufacturing quality management. It is well understood that in the near-term human oversight is still needed. While generative AI can accelerate many tasks, human review is still important to validate the results. Data governance practices will be crucial to ensure the accuracy and reliability of the data processed by generative AI. None of that should diminish manufacturing’s enthusiasm for a set of tools that can create revolutionary improvements in quality approaches. AI-driven solutions are not just a technological shift—it’s a strategic move that could redefine quality control in manufacturing.