How to Succeed in Manufacturing AI Compliance: A Trustworthy AI to Win?

As discussed in Part 1 In this series, manufacturers can gain an uncanny AI competitive advantage through defect detection, predictive maintenance, and automated asset management. But the power of AI goes beyond these use cases, supporting a whole new dimension of automation and insight, shares Lori Witzel, research director for analytics and data management at TIBCO.

Artificial intelligence (AI) is in the news, as are regulations for AI risk management. AI regulatory compliance will affect manufacturers sooner rather than later.

Through artificial intelligence and related technologies, manufacturers can have a complete, integrated, data-driven 360-degree view of all operations—from suppliers and supply chains, through equipment, processes, and manufacturing practices, to final product testing and customer satisfaction. The promise of Industry 4.0 has been fulfilled, and it is widening the gap between the leaders and the laggards.

However, the benefits of AI are no longer without risks. Increased adoption of AI across many sectors, including manufacturing, is leading to increased technological regulation. American manufacturers need to act now to prepare for the changing regulatory landscape.

Trustworthy AI is best practice

Building trust and transparency in AI is an essential best practice. It is also necessary to ensure compliance with current and future regulations.

A trustworthy AI is auditable, transparent, and explainable (with the risk of oversimplifying a complex subject). Explainable AI includes algorithms that clearly explain their decision-making processes. This interpretation ensures that humans can evaluate an AI-infused process, so that they can apply their own insights and opinions to the reasoning behind a decision made by the AI.

For example, an experienced operations manager may need to understand why some products that come through production are identified as defective and not others. If the AI ​​determines that a product in an image is defective, this is a possible use case for interpretation – the need for a human to be able to validate the decision. The AI ​​becomes interpretable when the location of the defect is marked visually, so that the person can see and verify which of the many visual features in the image represents the defect. This cannot be explained if the AI ​​only indicates that the image contains a defect but does not highlight the actual defect within the image.

Another example of manufacturing-specific risks, Mackenzie noticed it, is the potential for accidents and injuries due to the AI ​​interface between people and machines. If AI-implanted systems fail to keep a human in the loop — should interpretive best practices fail — equipment operators may not be able to provide the required override, increasing physical risks in applications using autonomous vehicles. Other risks to manufacturers, such as downsizing the supplier’s faulty AI, are also implications.

Explainable and transparent AI will enable data science teams to respond in ways that even the least technical workforce can understand. This is particularly useful for legacy manufacturing operations, which often find themselves under pressure from digital competitors.

See more: A Quick Guide to Intelligent Manufacturing

Trustworthy AI is based on reliable data

An example of the value of reliable data for manufacturing is Arkema, a €8 billion French specialty chemicals and advanced materials company. They make technical polymers, additives, resins and adhesives. The flow of data across domains of customers, vendors, and materials across the business has revolutionized it with their data-weave-like approach to data assets. Jean-Marc Vialati, Group Vice President of Global Supply Chain at Arkema, has led an enterprise-wide initiative that puts a common data framework into an ever-expanding list of products, ensuring that every system deployed is pulled from the main trusted data center.

The Arkema team now widely shares standardized and trusted data across the enterprise, enabling enhanced regulatory compliance, facilitating incremental growth through integration of data on M&A activity, and supporting impeccable customer-focused service. Arkema is an example that U.S. manufacturers can learn from as they seek advantage by using AI for supply chain optimization, anomaly detection, root cause analysis, key factor identification, yield improvement through large-scale pattern recognition, and predictive and educational maintenance via advanced equipment monitoring.

How to prepare for the changing AI regulatory landscape

As noted by McKinsey, manufacturers that use AI are vastly outperforming their counterparts that are lagging behind. The examples they cite lead to loss reductions of 20 to 40 percent while improving on-time delivery using an AI scheduling agent. But without preparing for AI transparency and auditability, these advantages may be lost due to regulatory risks. Although regulation of AI remains on a country-by-country basis, in many cases, and is in the draft stage worldwide, preparation for implementation according to compliance could include:

1. Data Fabric Architecture with Robust Master Data Management (MDM) for end-to-end management of data pipelines that feed manufacturing automation: Regulatory compliance means understanding not only the algorithms used but the data that has been used to train AI and machine learning (ML) models. Data texture provides a framework for achieving transparency as well as better results.

    • Discover and manage AI training data: Not only may data science teams use data from the enterprise, including IoT data, but they may also use publicly available datasets. Whether the data source is internal or external, data attribution, observability, and transparency in its use are essential components of regulatory compliance.
    • Discovery and management of personally identifiable information (PII): To ensure regulatory compliance with AI, the organization must understand whether there is personally identifiable information in any AI system the organization uses. A powerful mobile device management tool can help identify PII data in which systems and how PII is hidden or otherwise protected.

2. Data virtualization to help scale and reduce friction in preparing AI training data: The sheer volume of training data that machine learning and AI systems need requires flexible and scalable data prep processes. Data virtualization can reduce friction in preparing data by reducing the impact of data silos on scalability and access.

3. Basic and running algorithm audits: Identifying and documenting algorithms used across manufacturing automation and supply chain processes is an important measure toward the transparency needed for regulatory compliance.

    • Algorithm transparency and interpretability: An integrated platform approach to data analytics and data science will make identifying and documenting the algorithms used easier. It will also help ensure the transparency and interpretability of these algorithms – key aspects of AI compliance.
    • Trading Partner Documentation and Seller Algorithm: Manufacturers should also require business partners and technology vendors to document any algorithms that the manufacturer’s systems and processes may use. Boston Consulting GroupAmong other things, it recommends implementing a responsible AI framework that includes vendor management where a manufacturer may be liable for non-compliant AI provided by a business partner or vendor.

Just as the benefits of AI for manufacturers transcend silos and extend across the organization and its business partners, so too should preparations for the regulation of these technologies. Artificial intelligence can be pivotal in enabling manufacturers to leap ahead of the competition. As you prepare to make that leap, ensure you have governed and transparent AI processes in place – along with diverse stakeholder input – to be able to adapt to the changing regulatory landscape.

What AI compliance strategies are you implementing to adapt to the evolving regulatory landscape? Share with us on FacebookAnd TwitterAnd linkedin.

More about manufacturing

Leave a Reply

Your email address will not be published. Required fields are marked *