There’s One Problem AI Can’t Resolve: Implementation

Speaking, as we were last month, about artificial intelligence and its adoption into electronics design and manufacturing, we observed that a current obstacle to implementation is the use by vendors of customer data in order to build their models.

And while vendors insist the data are aggregated and anonymized, said customers, naturally, have been generally circumspect over the perceived cost of the lessons they have learned – often painstakingly – being used to enable competitors, not to mention ultimately paying those same vendors for the courtesy.

To that I will add the musings of Neil Thompson, who is the director of the future tech research project at MIT’s Computer Science and Artificial Intelligence Laboratory.

Thompson argues that AI systems must not just be capable of performing “human” tasks but also must overcome the costs of implementation, including redesigning processes and methodologies. “There are a lot of places where … humans are a more cost-efficient way,” he says.

Over the past couple years, via our PCB Chat podcast and webinars, our editors have spoken with a growing number of AI experts across the spectrum of the electronics supply chain. They include:

• Sebastian Schaal, cofounder of Luminovo, a Munich-based startup working on applying machine learning to electronics development and production processes;
• Leah Slaughter, then vice president of supply chain and artificial intelligence programs at a multinational EMS company, on how AI is being used to improve electronics supply chain management;
• Matthias Wagner, chief executive and cofounder of Flux, a developer of a browser-based PCB design platform and programmable simulator;
• Andrew Scheuermann, cofounder and CEO of Arch Systems, a developer of software tools that collect raw machine data and use predictive analytics to calculate manufacturing key performance indicators (KPIs);
• Tomide Adesanmi, founder of Circuit Mind, an AI-driven component selection and schematic capture platform;
• Arif Virani and Bart Piwowar, then chief operating officer and CTO, respectively, of Darwin AI, a manufacturer of innovative inspection systems (now acquired by Apple);
• Taylor Hogan, Cadence Design Systems; Kyle Miller, Ph.D., Zuken; and David Wiens, Siemens Digital Industries, major ECAD vendors that have invested in AI-driven features.

And see our interview this month with Sergiy Nesterenko, founder and CEO of Quilter, an AI software company whose goal is to accelerate hardware development by fully automating circuit board design.

The suggestions that AI can reduce time spent coding are well-accepted and, indeed, one would hope the effort that goes into setting constraints could be automated (assuming they are correctly prompted – precise communication is key to effective AI use) as well. But it all turns on specificity. As Scheuermann told us earlier this year, “What we have to do in manufacturing is combine these generic AIs that are being built with the specific data of each manufacturer’s problem, which needs to be kept private and secure only to them, so they can combine these things in a controlled environment and then they can get the best use of AI combining general AI and their specific data for their problems.” While Scheuermann was couching his opinion around manufacturing, it applies all the same to design.

As the business models continue to take shape, vendors will have to think long and hard about how to overcome user hesitancy. To that end, we are excited to announce our keynote speaker for PCB West in October is Charles Pfeil. Few understand the nature of printed circuit design tools better than Pfeil. He was the original product architect of Expedition PCB, and an inventor of Team PCB, XtremePCB, XtremeAR, and the Sketch Router. If anyone knows the nature of engineers when it comes to implementing new tools and features, it’s Pfeil.

Pfeil’s talk, “AI Roadmap for PCB Layout: New Opportunities for PCB Designers,” will home in on the impediments to adoption and how to overcome them. His solution involves balancing the growing capabilities of AI with the preservation of proprietary design knowledge and methodologies. (His talk will take place October 9 at 11 a.m. at the Santa Clara [CA] Convention Center.)

As PCEA continues to work on the industry roadmap for implementing AI, Pfeil’s talk is sure to spur thoughtful action toward a workable solution for a conundrum even AI hasn’t been able to solve.

P.S. Our thoughts are with the families and colleagues of George Dudnikov, an engineer and executive at several leading PCB fabricators, and Jim McElroy, the first CEO of iNEMI, both of whom passed away this summer.

Mike Buetow is president of PCEA (pcea.net); mike@pcea.net.