Embracing AI in aerospace manufacturing

Ask anyone in manufacturing what the future of the industry looks like and you’re likely to get one answer: automation and artificial intelligence (AI). While many have legitimate concerns about the technology – i.e., will it replace human workers? Does it have the potential to be dangerous? – it’s clear that to survive in the industry, manufacturers must shift their thinking toward automation and AI.

“AI is not going to replace you, but you’re going to be replaced by someone who knows AI,” says Sung Kim, chief technology officer at cloud software developer iBase-t.

It’s a blunt truth, but it’s true, especially in such a data-heavy, competitive, highly regulated industry as aerospace manufacturing. When it comes to processing and tracking all that data, AI is a necessity.

iBase-t’s flagship product, Solumina, is a manufacturing operations platform to help manufacturing companies, particularly in the aerospace & defense sector, simplify their operations, ensure precision, and standardize production processes through advanced technology. Toward the end of last year, the company introduced an AI element into the platform. Solumina AI includes a digital assistant, large language model, Industrial Internet of Things (IIoT) integration, and other features for process optimization.

A digital knowledge base for each company

The decision to pursue AI-based solutions was an easy one for iBase-t.

“The great minds in aerospace and defense manufacturing are retiring with the tribal knowledge they’ve gained for decades,” Kim says. “You don’t want to lose that data, and it’s going to take a lot of resources, time, and budget to retrain the younger workforce. So this is a really great time to put that domain knowledge into a scalable format, which is AI models.”

While Solumina AI leverages generative AI, it’s specifically designed to be trained on the user’s manufacturing data rather than public data, making it a knowledgeable manager for shop floor operations. It integrates with product lifecycle management (PLM) and enterprise resource management (ERP) software, as well as gathering data from operational technology (OT). These integrations allow it to track production steps, material origins, and everything else involved in the manufacturing process to create a detailed digital thread.

Many aerospace manufacturers are gravitating toward this type of vertical generative AI, Kim says, as well as the more active, autonomous agentic AI (which is designed to take actions, rather than generating outputs) working alongside human workers. “Alongside” is the key word, as even the most sophisticated AI requires human oversight.

“There’s always a human being involved, because at the end, we’re going to be accountable for any action taken, so the humans will have to approve and make it better and take it further,” Kim says.

Ensuring data security, accuracy

In addition to the fear of being replaced by AI, the other major concern among consumers is cybersecurity, which iBase-t takes very seriously. The company is working closely with regulations such as NIST 800 171 and International Traffic in Arms Regulations (ITAR) to ensure compliance. It implements data encryption in transit and at rest, and employs strict access control measures as well as regular audits. The company is also working to detect AI hallucinations, or incorrect answers, and ensure factual consistency.

Kim is confident people will become increasingly comfortable with AI – as with any emerging technology, there’s mistrust at first, but that lessens as the safety and efficacy of the technology is proven.

“We need to wait a little longer to see the maturity of the data handling on the AI side, and not just the maturity, but the transparency and visibility,” Kim says. “We have to know what’s going on so we can actually do better.”

The large scale of AI operating within a facility is a concern, he continues, because if a human being gives an incorrect answer, it likely impacts a few people or operations, but if AI gives a hallucinated answer, it can impact thousands of operations. That’s why the manufacturing industry must improve its factual consistency checking and monitoring techniques.

Autonomous technology, human-machine collaboration

The recently introduced Solumina AI model is just the beginning for iBase-t. Currently, the company is working with data intelligence, which uses an AI technique to harness the power of Solumina data and develop better data-driven decision-making capabilities. It’s also pursuing more agentic AI applications, such as small Solumina agents to act upon the data and decisions to improve shop floor efficiency and guide best practices. Kim also envisions training AI to continually update itself according to the latest regulatory policies and ensure the digital artifacts it creates are in compliance.

The development of AI technology is moving fast and could look drastically different in just a few months, Kim says. One of the keys to progress is increased data integration – with digital twins, with IIoT, allowing AI access to multiple record systems so it can accurately guide an agent or chain of agents. The dream of the manufacturing industry is the lights-out, smart factory, which we’re leaning toward as agentic AI integrates data to become more autonomous.

“I’ve seen a use case where, when the equipment is about to fail or in failure mode, (the AI agent) actually detects the root cause based on the data that equipment generates, as well as the past repair history or maintenance history, and points toward how to fix it,” Kim says.

He reiterates, however, that humans have an active role to play in the continued development and operation of AI, and that when AI takes over tasks such as data entry or troubleshooting, it frees up human brainpower for innovation and creation. Manufacturing will become a team effort between humans and AI.

“We all have to monitor what’s going on,” Kim says. “We have to be auditors of what’s happening in the AI space and to make sure we’re using it correctly, to enhance the humans’ ability to carry out tasks so we can actually do more meaningful stuff at the end of the day.”

iBase-t
https://ibaset.com

About the author: Clare Scott is associate editor for GIE Media’s Manufacturing Group of publications. She can be reached at CScott@GIE.net.