Artificial Intelligence and Manufacturing Operations, What Might The Future Look Like?

As we transition from the first three Industrial Revolutions—characterized by steam engines, mass production, and computerization—we find ourselves in the midst of the Fourth Industrial Revolution, or Industry 4.0. This era is marked by Artificial Intelligence in the form of Generative Pretrained Transformers (GPT), which demonstrate abilities we could only have dreamt of a few years ago. Although these technologies have yet to fully manifest in aerospace manufacturing, they hold the promise of transformative change. So, what will the future look like? My view is that it will be entirely different yet exactly the same.

What is Artificial Intelligence and How Did We Get Here?

Let’s start with what is meant by Artificial Intelligence. Artificial intelligence itself is an interesting and potentially misleading term. An AI algorithm is using complex math to produce the most likely output based on an input, typically a digital image or text. In truth, it is closer to applied statistics than to a form of intelligence. Nevertheless, the term “AI” has been accepted and is far from charmless.

This technology is based on a mathematical representation of a neuron first implemented in the 1950s. The advent of the Graphical Processing Unit (GPU), initially designed for video game graphics, provided the computational power necessary for AI development. Yes, it is true - whilst playing your favorite first-person video game, you were unknowingly laying the groundwork for AI. We now have the basic structure of a neural network and the specialized hardware needed to quickly run the required calculations. It took a while to realize this, with the first glimpse of this future in 2012 with Alexnet.

The convergence of internet infrastructure, digitization, and academic research has led to today’s AI capabilities. To get the internet running, we needed rooms filled with computers, which improved as online activity grew. Eventually, we had excess computing capacity, leading to the creation of cloud computing. These computers were versatile and capable of various tasks. Simultaneously, researchers were using large computing power and neural networks for pattern recognition, utilizing vast amounts of digital data. With a few improvements in hardware and neural network architecture, we arrive at today’s capabilities.

Right now, in aerospace manufacturing, we already connect all objects of any significance in the factory to the Internet of Things. 

That gives us streams of data from up and down the supply chain and manufacturing process. So, here we are in the future, and what does this all mean?

The future will be entirely different yet exactly the same.

Let’s start with things that will be the same. Highly optimized manufacturing processes won’t change. Heavily loaded cylindrical components will continue to be made from wrought bar stock on a lathe. With a high enough batch size, the lathe will be tended by a robot, and any lathe will be CNC controlled. But, otherwise, the process will be the same as today. A manufacturing engineer will still lay out the manufacturing plan and produce instructions and G-code. For the majority of manufacturing operators, activities will remain the same too. They will monitor the process and intervene when required. Automation of “everything” will remain out of reach.

"Companies and Individuals Will need to be Adaptable as New Technologies Emerge and Old Ones Become Obsolete"

On the entirely different side of the ledger, to operate and maintain an AI-enabled machining center will require much more software and electronics skills. Whilst the manufacturing engineer is laying out the process, AI algorithms can analyze vast amounts of data to identify inefficiencies and suggest improvements. For instance, machine learning models can predict equipment failures before they happen, allowing for preventive maintenance that minimizes downtime. As AI systems become more sophisticated, they will act as decisionsupport tools for engineers and operators. For example, AI could suggest the most efficient routing of materials through a factory or recommend adjustments to machine settings in real-time. However, AI cannot run the entire process alone. The manufacturing engineer will need to check the work of their AI assistant, who makes the occasional mistake.

AI can also revolutionize quality assurance. Advanced image recognition can identify defects in products faster and more accurately than the human eye. This not only improves the quality of the output but also reduces the costs associated with recalls and returns. At Moog Inc., we are making complex metal parts through a metal additive manufacturing process called Laser Powder Bed Fusion. It allows engineers to create more complex parts than ever before and promises to increase the performance of systems using additively manufactured components or assemblies. Now, we’re deploying Artificial Intelligence to help inspect the quality of work. We teamed up with the University at Buffalo New York State Center of Excellence in Materials Informatics (CMI) to apply convolutional neural networks to create a highly trained computer algorithm that can recognize high quality additive manufactured parts and reject the lower quality ones. This helps technicians inspecting the parts welded by Laser Powder Bed Fusion do their jobs more effectively and efficiently.

So how should we be preparing for this future?

Let’s first remember that AI has no “intention” and doesn’t have any needs or wants. It can usefully be compared with a power tool. If you are going to drill a hole in the wall, an electric power tool will save time and effort. The same will be true for many computer-based activities such as writing lines of code or responding to emails. There will be an ongoing need for people to direct AI in the workplace. There are also ethical and social implications. Companies will need to address these issues proactively, ensuring that technology serves to augment human labor rather than replace it.

As with other transformative periods of history, the only constant is change. Companies and individuals will need to be adaptable as new technologies emerge and old ones become obsolete. Continuous learning will be increasingly important. The future will require a more flexible approach, rather than sheer memorization. Educational institutions and companies will need to develop training programs that teach people how to learn effectively. Finally, we will all need to keep our minds open and our flexibility to the maximum.