Beyond the Buzz: From Theory to the Factory Floor - What Manufacturers Must Know About AI’s Journey

#Industry40 #FactoryOfTheFuture #ArtificialIntelligence #AI #AppliedAI #IndustrialAI #IITBombay #AsimTewari #SmartManufacturing #AIinManufacturing #ManufacturingTech

"Since 2010, AI has been growing exponentially. Every few weeks, it becomes significantly more powerful. This is not linear progress—it’s geometric. "- Prof. Asim Tewari, Department of Mechanical Engineering, IIT Bombay.

March 2025 : Artificial Intelligence is not new—but the world’s ability to understand, train, and scale it has changed dramatically in the last 15 years. In the first episode of our exclusive Masterclass on Applied AI in Manufacturing, Prof. Asim Tewari of IIT Bombay walks us through the remarkable evolution of AI, debunking popular myths and revealing the mathematical and technological breakthroughs that are powering today’s smart factories.

With over 20 years of research and practical experience in machine learning and deep learning, Prof. Tewari is a distinguished academic at the Center for Machine Intelligence & Data Science (C-MInDS) and the Department of Mechanical Engineering at IIT Bombay. His perspective, grounded in deep technical expertise and close industry collaboration, challenges conventional thinking.

“There’s a huge gap in understanding what AI truly is. Some think it’s all hype; others believe it can do everything. Both views are wrong,” Prof. Tewari says.

He takes us back to the 1950s, when AI first emerged—not as a tangible technology, but as a concept popularized in science fiction. Decades passed with minimal progress until the late 1970s and 1980s, when machine learning (ML) gave AI a new path: letting machines learn from data, rather than being explicitly programmed. This marked the first significant leap.

But even ML had its limits. As Prof. Tewari explains, early machine learning models would show promise with small datasets but plateau with larger ones. Their performance couldn't rival human intelligence, and many believed AI had hit a ceiling.

That’s when a mathematical breakthrough rewrote the rules.

In 1989, the Universal Approximation Theorem proved that it was theoretically possible for certain mathematical functions—neural networks—to model any real-world phenomenon to any degree of precision. It was a revelation, but one that seemed impossible to apply. The models were too complex to train on existing hardware. AI remained largely a theoretical promise.

The game-changer came not from academia, but from gaming.

Around 2010, researchers realized that Graphics Processing Units (GPUs)—originally designed for rendering video game graphics—were ideally suited to the matrix-heavy computations required to train deep learning models. “AI is nothing but math,” Prof. Tewari reminds us. “And GPUs were perfect for doing that math fast.”

This convergence sparked what he calls the Cambrian Explosion of AI—a reference to the period in Earth’s history when life diversified at an unprecedented rate. In AI’s case, the explosion began with the training of the first deep neural networks, like AlexNet, which demonstrated the true power of deep learning.

Companies like Nvidia, once known for gaming chips, suddenly found themselves at the heart of the AI revolution. Today, Nvidia is the world’s largest manufacturer of AI hardware, driving progress at a breathtaking pace.

“Since 2010, AI has been growing exponentially. Every few weeks, it becomes significantly more powerful. This is not linear progress—it’s geometric,” says Prof. Tewari.

This explosion of capability is transforming industries—including manufacturing—at a foundational level. Deep learning systems can now process massive amounts of sensor data, identify anomalies, and even predict equipment failure with superhuman accuracy.

Yet, Prof. Tewari offers a word of caution: while AI has grown exponentially, its limits are governed by two factors—data and compute power. “Give it more, and it grows. But we don’t have infinite data or infinite power. That’s what keeps us in check.”

In upcoming episodes, Prof. Tewari will delve into the practical applications of AI in manufacturing, from intelligent inspection systems to predictive maintenance, and how enterprises can responsibly and effectively adopt these technologies.

Pro MFG Media Newsletter

Get regular manufacturing insights that matter and enable business.