The State of Artificial Intelligence in 2025

The State of Artificial Intelligence in 2025

While 2023 marked the world’s formal introduction to AI with the debut of ChatGPT, 2024 was defined by the exploration of its limitless potential and capabilities, as many early adopters dove in headfirst. Now, as we head into 2025, AI aspirations are expected to be slightly tempered, with manufacturers adopting a more practical, strategic approach to implementation, driven by increased access to knowledge and resources to guide their AI journey. I also believe that 2025 will mark the year when more early adopters will begin to realize tangible value from their AI investments.

So, what does that mean for asset-heavy industries like manufacturing? I’m breaking down the trends and predictions that I expect will have the biggest impact in 2025 – from the rise of industrial agents to enhanced productivity, and how AI will help address the ongoing talent gap.

2025: The year of industrial AI agents

In 2025, the most successful industrial companies will move away from generalized language models to implement tailored solutions like industrial AI agents. Industrial AI agents utilize algorithms and data models specifically optimized for the patterns and anomalies typical in a particular domain’s tasks. They offer more accurate and relevant guidance and can also be scaled to accommodate growing data and operational complexity, improve decision-making processes and lead to higher productivity, safety and operational efficiency. This year, AI needs to meet users where they already are (hint: in their workflows), rather than moving workflows into a new system. By utilizing industrial agents, users won’t need to prompt the agent to do everything and workflows in general will become more efficient, thus improving productivity, safety and operational efficiency.

Organizations like Aker BP are already demonstrating the transformative power of domain-specific AI agents and illustrate how early adopters will gain a competitive edge. The company implemented a Document Parser AI Agent fine-tuned to understand unstructured technical documentation and its equipment hierarchy. This helped streamline equipment management processes, saving thousands of hours previously spent on manual data entry and enabling experts to focus on high-value business problems.

Manufacturers will overcome the fear that AI is taking jobs, and instead embrace its potential

AI, and specifically industrial AI agents, holds the potential to skyrocket a manufacturer’s productivity. However, to unlock this potential, we need to shift our mindset from viewing increased AI efficiency as simply a matter of increased personal efficiency. Let’s walk through a specific example. In asset-heavy industries root cause analysis (RCA) is a critical process that typically takes several weeks, if not months. I’ve worked with companies to introduce an RCA-specific industrial agent that reduces several of the major, time-intensive aspects of the RCA process to just 30 minutes. By programming the agent to take the incident and provide it with all relevant information across multiple data types, it can integrate, contextualize and perform an initial fault tree assessment based on the incident and the data we’ve collected. The subject matter expert (SME) can then use that assessment to finalize the RCA.

While a common concern is that we’re replacing humans with industrial agents to complete major steps in this workflow, the most successful professionals will recognize that RCAs are such an intensive process that they currently aren’t able to complete nearly as many without an agent. It’s not about replacing a human’s job, it’s about streamlining efficiencies so significantly that we actually unlock more opportunities. In addition, it frees up the SME to focus on higher-value, more rewarding tasks. In 2025, I’m optimistic that we’ll overcome our fear that AI is taking jobs, and adopt a more progressive approach that embraces its potential to significantly improve production, uptime and throughput, while simultaneously enhancing employee experience and morale.

Data preservation is key to navigating the talent gap

Manufacturing job shortages are expected to continue into the new year, making AI implementation even more critical to maintaining a consistent workflow.

Since the pandemic, the manufacturing industry has lost around 1.4 million jobs and as of January 2024 there were still 622,000 total manufacturing job openings that remain unfilled. The most successful manufacturers will make the preservation of historical data a key component of their 2025 AI integration strategy, in an effort to navigate the talent gap and stabilize operations. Companies like Celanese implemented AI technology early on to accelerate their digital transformation journey and power its Digital Plant of the Future, which turns their operations into digital factories to empower their employees with new tools that allows them to do better work more easily. This ultimately improved their tracking and analysis of historical data, which minimizes disruption and allows for better knowledge retention – a capability that will be essential as the industry braces for ongoing workforce transitions in 2025.

From the growth of industrial agents to leveraging AI to navigate the talent gap, 2025 is going to be another big year. Overall, I expect to see wider adoption as organizations witness AI-driven benefits and growth from their competitors across industries and realize that they must catch up or risk falling behind. As we continue to move towards a world where asset-heavy organizations are embracing AI, the manufacturing industry should urgently get on board to remain competitive and accelerate future success.

About the author

This article was written by Jason Schern, Field CTO at Cognite.

 

 

Related articles:

• What does it take to work with Generative AI Models
• Enhancing Manufacturing Processes with AI and Low-Code: An Essential Guide for Modern Manufacturers