Podcast: The Practical Guide to AI on the Factory Floor

That's a huge shift, especially for manufacturing. But to really grasp how AI will achieve this on the factory floor, we first need to get clear on what we mean by 'AI'. It's not just one thing. You've probably heard terms like 'Generative AI' and 'Agentic AI' thrown around, alongside the machine learning many are already familiar with. Let's break down what these different types are and how they fit into the manufacturing picture.

So the big terms you keep hearing are Generative AI and Agentic AI. And it's important to know they're not the same thing.

Before these newer types hit the headlines, many in manufacturing were already using AI, specifically machine learning. Think of machine learning as being focused mostly on prediction – it looks at data, finds patterns, and uses rules we give it to solve specific problems. You see it in predictive analytics, understanding language—that's natural language processing or NLP—and even some basic autonomous systems.

Then you have Generative AI. It also learns from data, but its main job is to create something new based on those patterns. This is the AI that writes text, generates images, maybe even composes music, or designs product models. Think ChatGPT, Claude, DALL-E, Midjourney—those tools that exploded onto the scene? That's Generative AI. It's great for mimicking human-like creative work and can be a real time-saver.

Now, Agentic AI is built to interact with the world around it, make decisions, and actually do things to reach a goal—all by itself, no human constantly telling it what to do. That's a key difference from Generative AI, which usually needs a human prompt to get started.

For Agentic AI, think about a self-driving car figuring out traffic based on its sensors, or a robot arm in a warehouse picking and sorting items on its own, or even a smart assistant that manages your calendar without you asking. These systems use sensors to 'see' or 'feel' their surroundings, algorithms to 'think,' and actuators to 'act'.

The core idea with Agentic AI is that it has a goal. It doesn't just react; it considers its objectives and makes choices to get there. This is the type of AI people sometimes joke about taking over—because it can act independently. That independence naturally brings up serious questions about ethics and accountability. Like, who's responsible if it messes up? How do we ensure it stays aligned with what we want it to do? Developers are definitely grappling with this, building in safeguards to keep things on track.