FactoryTwin® CEO Vivek Saxena Shares Perspective on “AI in Manufacturing: Myth vs. Reality”

Hopkins, MN —November 2025— Dr. Vivek Saxena, CEO of FactoryTwin and a manufacturing practitioner with three decades of industry experience, has shared a three-part perspective titled “AI in Manufacturing: Myth vs. Reality,” examining common misconceptions and practical realities surrounding the use of artificial intelligence in manufacturing.

Drawing from decades of work in manufacturing systems, the series addresses why AI adoption in manufacturing continues to lag other enterprise functions and why much of the current focus on generative AI does not align with shop-floor realities.

In the first part of the series, Dr. Saxena emphasizes that manufacturing is “way more complex than any other enterprise function,” citing high dimensionality, strong inter-dependency of variables, and low data quality that is often unappreciated by non-manufacturing professionals. He distinguishes between two main flavors of AI: Analytical AI, which is well grounded in mathematics and works on structured data with high reliability and predictability, and Generative AI, which is creative but not always reliable. He notes that claims of fully autonomous factories driven by general AI models are years, if not decades, too early.

In the second part, Dr. Saxena explains why, in manufacturing, mathematically sound analytical AI and machine learning models should be used only when they are proven to work, rather than relying on LLMs or generative AI. While GenAI has interesting niche applications, he states that it is not suitable for complex manufacturing environments and challenges vendors and users to demonstrate significant improvements in shop KPIs using only LLMs. He also notes that the hype around LLMs and GenAI is beginning to fade as their limitations become more widely acknowledged.

The third part contrasts the futurist vision of self-adjusting autonomous factories with current industry adoption data. Citing a 2025 McKinsey AI usage survey, Dr. Saxena notes that only five to six percent of companies have tried using AI in manufacturing, compared to about thirty percent in functions such as marketing and IT. He attributes this gap to the complexity and integrity challenges of manufacturing data, the repeated focus on narrow use cases such as preventive maintenance and computer vision–aided quality control, structural barriers similar to the long-standing “hands problem” in robotics, and the mission-critical nature of manufacturing activities such as scheduling.

Despite these challenges, Dr. Saxena highlights that there are impactful and targeted use cases for AI in manufacturing. FactoryTwin, in collaboration with the University of Texas at Dallas and the University of Minnesota, is working on agentic AI solutions that are more generic and reliable, while maintaining a healthy skepticism of the applicability of generative AI in manufacturing.

The “AI in Manufacturing: Myth vs. Reality” perspective reflects Dr. Saxena’s three decades of experience in manufacturing and his continued focus on applying reliable, mathematically grounded approaches to complex factory problems.

About FactoryTwin

FactoryTwin is a factory intelligence platform powered by digital twin and AI technologies, purpose-built for small and mid-sized manufacturers. By connecting & curating IT, OT, offline & tribal knowledge into a unified view, we transform disconnected data into predictive and prescriptive insights that improve revenue, profitability, cash flow, and on-time delivery by double-digits. Backed by NSF-funded R&D and designed for rapid, low-disruption implementations, FactoryTwin has already helped dozens of manufacturing teams move from data to decisions.

Learn more: www.factory-twin.com