AI Adoption in Manufacturing: Challenges and Opportunities
Introduction
- Speaker: Nicole Rathy, Project Manager, AI for Manufacturing Canada
- Moderator: Jay Myers, CEO of Engine Canada, Business Economist specializing in industrial and technological change
- Panelists:
- Dana Watanabe, Data and Analytics Manager, Linamar
- Aaron Grant, Machine Learning Solutions Architect, Impact AI
- Format: Q&A period at the end; submit questions through the chat
Jay Myers' Introduction
- Focus on building advanced manufacturing solutions (automation, AI, ML)
- Addressing excessive hype around AI
- Notable global challenge: 75-80% of advanced manufacturing transformation projects fail
- Canadian manufacturers: 60% did not achieve business objectives from their investments in tech
- Essential to manage technology productively to achieve business goals
Key Focus Areas
- **Strategic Questions for AI Implementation: **
- Where in a process can AI improve?
- What processes add customer value?
- Requirements for successful AI implementation (necessary data, data quality)
- **Potential lies across manufacturing domains: **
- Logistics, risk assessment, systems optimization, financial planning, rapid prototyping, embedding autonomy
Dana Watanabe on Linamar's AI Implementation
- Role: Data and Analytics Manager, Strategic AI Council member
- Approach: Decentralized; focus on upskilling employees in data literacy and AI
- Training: Emphasis on data quality, continuous improvement cycles (e.g., PDCA - Plan Do Check Act)
- Projects: 150 trained workers, 25 successful AI projects in 2022
- Challenges: Variation in data maturity across facilities, facilitation of collaboration among facilities
Aaron Grant on AI Use Cases at Impact AI
- Focus: Helping small manufacturers improve productivity through AI
- Use Case Example: Enhancing laser welded steel sheet manufacturing using AI to improve weld inspection processes
- Approach: Emphasize on harnessing existing sensor data for predictive analysis
- Outcome: Increased productivity with existing assets, better defect detection
- Challenges: Variability in data quality, resistance to changing from well-established processes
Common Challenges in AI Adoption
Data Management
- Garbage in, garbage out principle: Importance of data structuring
- Manufacturers may have too much data, leading to potential overwhelming scenarios
- Systems of prediction must be intelligently designed
Cultural Adaptation
- Adapting processes to AI, getting buy-in from different organizational levels (workers vs management)
- Training and making data management an asset
- Addressing skepticism from both management and operational staff
Small vs Large Companies
- AI solutions are scalable irrespective of company size
- Smaller companies can use AI with targeted, well-defined objectives
- Leveraging Open Source models and accessible programming (Python)
Guidelines and Standards for AI
- Developing Best Practices: Focusing on documenting guidelines for data quality and AI integration
- Customization: Addressing specific problems with targeted AI solutions, involves significant integration work
- Formalizing Cybersecurity Procedures: Essential for any data-driven AI project, varies by data sensitivity and application
Practical Examples and Success Stories
- Image recognition for quality inspection
- Reducing health risks and improving safety with automated heavy lifting solutions
- Enhanced defect detection in manufacturing process
Wrap-up & Closing Thoughts
- Importance of aligning technological solutions with business goals
- Integrating technological capability with manufacturing needs
- Collaborative mindset between technologists and manufacturers essential for successful AI implementation
Upcoming Events
- Check AI for Manufacturing Canada and NGEN for more webinars
- Register for upcoming events in March
End Note: Thanks to all participants and panelists for their insights and contributions.