In this episode of the True North Podcast, IpX Director of MBE, Max Gravel and Senior Technical Program Manager, ASML, Arnaud Hubaux discuss the challenges of maintaining the physics model in AI/ML with rapidly changing datasets.

Arnaud gives a detailed review of how ASML uses AI to deliver greater value through new products for their customers and to accelerate manufacturing processes by qualifying high performing parts. Max and Arnaud discuss ASML’s unique approach to machine learning solutions with customers and how AI optimizes production and performance. Hear Arnaud break down:

• The challenges and necessity of explainability in AI/ML — where a causal clarification is needed between the observed effects and what is really happening physically in order to improve future design
• Ensuring continual guaranteed performance upon installation and after deployment of their machines to manage drift and ensure dataset accuracy
• The challenge and criticality of safety for people and materials first
• Defining and maintaining baselines consisting of thousands of parameters between both design and production
• Understanding the change impact to physics models: how to perform validation and verifications activities on the physics models before changes are applied to the customer product in order to mitigate risks with machine behavior.

In this second part episode of the True North Podcast, IpX Director of MBE, Max Gravel and Senior Technical Program Manager, ASML, Arnaud Hubaux continue their discussion on the challenges of maintaining the physics model in AI/ML with rapidly changing datasets.  Arnaud describes the need for an enterprise-level AI/ML framework that considers the complete lifecycle change to expand beyond niche use-cases and realize the dream of a fully automated product line. Max and Arnaud discuss where to start when implementing AI and the role of CM in maturing the adoption of AI in a manufacturing environment. Hear more on:

• The need for the impact analysis to stretch beyond the walls of the company into the environment where it will be running.
• The perceived unpredictability of AI/ML and the reluctance to deploy in a manufacturing environment.
• The importance of building in safety mechanisms to prevent financial loss in case of AI/ML failure.
• How CM establishes gates and checks when releasing a change to make sure the proper qualifications are implemented.
• AI/ML sustainability and the need to look beyond short-term benefits into the complete lifecycle of a product.