The 28th Annual ARC Industry Forum 2024 kicked off this week with a sharp focus on the ways that artificial intelligence (AI) is being applied in industry, with an emphasis on how AI can augment modeling processes and scale asset productivity and efficiency. Andy Chatha, CEO at ARC Advisory Group, said “we are all here for two goals: make new connections and learn from each other.”
Keynote highlights
The first speaker at Tuesday’s keynote was Wade Maxwell, vice president of engineering at ExxonMobil. He focused largely on the technologies behind Exxon’s energy transition, such a carbon capture and sequestration, hydrogen, low emissions fuels from agricultural waste, advanced plastic recycling and reuse, lithium extraction and methane detection equipment.
Exxon has also been active in developing artificial intelligence (AI) applications. The company has deployed AI for subsurface modeling, characterization and optimization and optimizing injection science for carbon sequestration as part of its low carbon solution. In Exxon’s own manufacturing environments, AI is used to optimize early anomaly detection on production assets and to predict reliability events.
“It's going to liberate and continue to liberate our people from spreadsheets, from data wrangling. It's going to hydrate the data that they currently spend a lot of time even finding and manipulating in insights, better faster decisions to put on a closed loop. You can take the human out of that. You can improve the cycle time, you can improve the quality, you can make better decisions, and you get a more direct line of sight to the creation and delivery of value,” Maxwell said.
He also highlighted that component and equipment interoperability is a key enabler for the energy transition. “We continue to advocate for open standards that enable interoperability between vendor solutions,” he added.
Michael Carroll, vice president of innovation, Georgia-Pacific, also focused on artificial intelligence in his keynote address, particularly generative AI (GenAI). He predicted that in 2024, we have entered the year of GenAI.
“What's generative AI for? It allows you to communicate with technology, technology to communicate with you and for you to harness and harvest, a cumulative knowledge that exists in your company,” Carroll said.
Industrial AI panel discussion highlights
The industrial AI executive panel was led by Colin Masson, director of research, ARC Advisory Group, who focuses on AI technology on industrial solutions. As well as introducing the panelists, he gave a preview of a new ARC technology leadership series, starting with a report called “Widening the Digital Divide—How Leaders are Embracing Industrial AI.”
The following at five important takeaways from the panel discussion.
1. Manufacturing skill sets are highly nuanced and challenging for AI to learn.
“The first thing that I learned is that manufacturing and logistics, industrial folks that make and move things, it really runs on skills, and that these skills, first of all, they take a long time to learn,” said Kence Anderson, CEO of Composabl. “These skills are very, very, very, very nuanced and no single algorithm out there, no AI model out there is going to be able to reproduce that, and that’s when I really started thinking about what I call intelligent autonomous agency.”
“I like to think of AI in two different ways: it tells you what is going on, what's happening, or AI that tells you what to do. I think it's important to understand that there's two sides of it,” he added.
2. AI-enabled decision making will take its seat at business meeting tables.
“We have three basic characteristics of AI that we can apply,” those being infused products, integrated services, and connected or augmented intellect, said Rashesh Mody, EVP business strategy and realization at Aveva. “One which is becoming very, very popular the last few years is about intellect. Can we augment more and more decision process ability to customers, giving them more guidance, more skilled management, so that they can improve the operations efficiencies in every aspect.”
3. Industrial AI scalability is here.
“Foundational models will significantly change the usage of artificial intelligence for two reasons. They will be much more accurate because someone has to put a lot of effort into that one model they often use later on,” said Axel Lorenz, CEO for process automation at Siemens AG. “And number two, they will be scalable.”
4. AI will need to correlate many different data modalities, depending on the industry and application.
“Another class or application [of AI] is generation of content, like building failure modes and effects analysis automatically, thereby reducing the amount of time,” said Jayant Kalagnanam, director of AI applications at IBM Research. “The other new class of AI applications is really targeting time series data, so, unlike language, time series is quite different in the sense that it's not a single channel data source, but multiple sensors at a time. So part of our focus has been to build new transformer architectures, which go beyond language to handle the multi-channel nature of time series, to actually do self-supervision to learn representations.”
5. Hyper-scalers like Microsoft are all in on industrial AI.
“How important is AI to you as an organization? Well, it’s in every application that we’re creating at the moment and bringing to market,” said Simon Floyd, general manager of manufacturing and mobility for Americas at Microsoft. “We are all in when it comes to AI and we believe that now is the right time.”
“I think it's a really exciting time actually, to be in this industry, because [generative AI] is the one technology I think everybody can understand how it's going to work without it being too challenging for many traditional processes. And so, I'm really, really excited myself actually to be part of this industry and helping other organizations realize the different ways that they can be used in something which is very modern, but not forget the investments that they've already made in time series analytics or whether it be machine learning for computer vision applications.”
The event welcomed close to 800 attendees, including more than 100 international attendees from 15 different countries, 195 speakers and panelists, more than 200 companies, and 60 media and other industry attendees.
