Issue link: https://iconnect007.uberflip.com/i/1511625
50 PCB007 MAGAZINE I NOVEMBER 2023 sary data out of the equipment in a consistent way. at doesn't exist in our industry as well as it could. Second, it's not necessarily com- puting horsepower, but more of an infrastruc- ture design issue. We're dealing with quite a bit of real-time data that needs to be transported over networks, stored, and processed. In other words, it's designing infrastructure that can han- dle the future needs of AI computing. e last piece is cloud integration, where much of AI is built. at starts moving us away from the phys- ical component, and into looking at how we can start processing the data at scale and building data sys- tems in a virtual cloud archi- tecture. ese are challenging things to put in place because of the overall cost and knowledge gaps that exist. Of course, when you are gath- ering and analyzing data, you're operating post-facto or in a reactive way. When it comes to registration, that tends to be more of a predictive, proactive activity— planning ahead for the variations and such. How do we move from one to the other? AI is a learning tool. We need to collect data on how individual companies' systems and pro- cesses operate. As more information is learned and interpreted, the decisions will become bet- ter over time. e predictive ability comes aer we're able to store, catalog, understand, and then predict where that next similar job will be from a scaling point of view. It's important to remember that the power of AI comes from the ability to take in so many vari- able factors—which we human beings are chal- lenged to compute—and put that all into a model. With that model, now we're able to take, say, 50 data points that were independent and compute them for the next job to get a recommendation, such as "is is where we need to be for scaling factors based on all the different permutations of those 50 data points. at's AI interpreting over 100,000 different permutations." Over time, those data points become more numerous and converge on good solutions. at's how AI could be impactful for our future. Are the tools available to provide this sort of predictive functionality? AI models generally need some research and engineering; the tools must be developed. As part of our Factory 5.0™ modernization steps, we're looking at standardizing data, data ware- housing, and ways to integrate with each one of our equipment manufacturers. It all must be custom built for our industry because there cur- rently is no model out there that understands how our industry operates. ere's an opportu- nity here, but there are chal- lenges before it can truly cre- ate impactful value. We'll have to overcome those challenges to translate that into something impactful in our manufacturing lines, creat- ing more advanced manufacturing processes as a result. I think it will take time, collaboration, and a lot of resources and knowledge from peo- ple to truly make it a part of our industry. Do you see the transition over time to AI techniques as mandatory? Like anything else, we all need to compete in the industry. We must stay relevant, and there are companies looking to advance their technol- ogy more assertively than others, whether by need or desire. Eventually, just like in the rest of the world, AI will become an integral piece of the ecosystem, but it's just a piece of the total ecosystem. ere will always be people; we still must exist. ere will be engineers and produc- tion work staff. We just have to find the middle ground, the collaboration. How can we be bet- ter at delivering what we're trying to build here with the tools we have? Agreed, good thoughts. Thank you, sir. ank you, Nolan. PCB007 AI is a learning tool.