velera How Credit Union Innovation Can Drive Gen Z Engagement May 2024 Banner

AI 1.0 Is Already Accelerating Industry 4.0’s Innovation Pace

Artificial intelligence (AI) is giving industry 4.0 a shot in the arm. 

The bleeding edge computing innovation is driving future-fit efficiencies within historically fragmented and labor-intensive industries, slotting productively into the current industrial era of connectivity, advanced analytics and automation. 

“The Industrial 4.0 revolution was based on digitization, and it’s starting to mature now,” Prateek Kathpal, president and CEO of SymphonyAI Industrial, told PYMNTS for the PYMNTS “Matchmakers” series. 

“The maturity of these technologies is also making it more reliable, powerful and cost effective [to] leverage these industry 4.0 solutions like IoT (internet of things), big data analytics and AI,” Kathpal said. 

Large language modelsv(LLMs), especially those purpose-built for the industrial sector, are set to accelerate digital transformations even further, bringing forth a new era of unprecedented advancements. 

And in light of the ongoing disruptions still being felt as a result of the COVID-19 pandemic, Kathpal noted that AI has emerged as a key enabler in responding nimbly and adeptly to disruptions.

“AI can help organizations respond to disruptions by helping in several ways, whether it is reducing your waste, whether it is your demand forecasting, or even optimizing your resource usage,” he explained. 

Powered by other still-nascent advances, industrial capabilities are being revolutionized across nearly every internal and external operational process.

A Revolutionary Convergence of Digital Technologies

The improved reliability and cost-effectiveness of various industrial technologies made it an opportune time for SymphonyAI to launch its own industrial LLM solution.

“Artificial intelligence is based on data. The more data you have, the better data you have, the cleaner data you have, then the better solution or a better output you’re going to get,” Kathpal said. “And the better and more relevant the output, the more actionable insights for better decision making you’re provided with.” 

Key use cases include supply chain optimization, demand forecasting and identifying inefficiencies in production, he added, as well as predictive maintenance analytics that enable the prediction of machine failures and optimized production schedules.

Crucial to success is the fact that industry-specific LLMs must process vast amounts of industry-specific data. 

But, as Kathpal explained, the outputs of even a highly complex industrial sector-trained AI model are not themselves overly complicated. 

“You’re not getting a very complicated analysis, you’re getting them in a very human, very readable format,” he said. 

Challenges and Unexpected Wins

Still, while AI outputs may be easy to understand, integrating AI solutions into existing tech stacks to supplement existing workflows and processes is a little more challenging, although not as much as one might think. 

Potential frictions may arise from user adoption challenges, particularly employee resistance to unfamiliar AI technologies. Education and communication play a crucial role in addressing these concerns.

“You need to integrate AI solutions with your existing legacy processes, and also integrate it with your upstream and downstream processes to get the full value out of the solution,” said Kathpal, emphasizing the need for real-time application integration. “But LLMs are going to help your employees become more productive as a part of their regular day-to-day job processes.”

He added that one of the “biggest challenges with any LLM or deep learning AI solution” is the data. 

“Data quality is the most important thing, and when building an LLM, the challenge is to anonymize that data,” Kathpal said. 

SymphonyAI’s Industrial LLM alone was trained on 1.2 billion tokens and 3 trillion data points.

Addressing security and compliance concerns, Kathpal highlighted the importance of hybrid or fully cloud approaches to mitigate data breach risks. Ensuring high-quality, industry-specific data for training is crucial for accuracy and precision, while regularly updating and monitoring the model, along with implementing a human loop for validation, helps enhance accuracy and reliability.

“One thing that’s very clear, is that every industrial player who is digitizing their processes is looking to capture efficiencies as a result of that modernization,” Kathpal said. 

Moving from a time-based to a predictive maintenance scheduling is one area he flags as a low-hanging fruit. 

Looking ahead, Kathpal said he envisions a future where AI, IoT, and the metaverse converge to improve real-time monitoring, optimize industrial operations and focus on sustainability in manufacturing and production. He emphasized the ongoing refinement of SymphonyAI’s LLM and platform, integrating customer feedback and expanding AI capabilities to cater to more industrial use cases.

“We are at the very beginning of AI today, and the real value of AI shines when it is integrated into a real-time process,” he said. 

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.