BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Three Ways To Overcome Industrial AI Roadblocks

Forbes Technology Council

Senior Vice President and General Manager, Artificial Intelligence of Things (AIoT) Solutions at AspenTech.

AI isn’t just becoming a disruptive force in the oil/gas and energy industries. It is a disruptive force. And its role in the industrial sector will only continue to evolve in 2022 and beyond.

My company, AspenTech, recently commissioned an independent survey that highlights the extent to which AI embedded in fit-for-purpose industrial applications, or "industrial AI," has already begun to (re)shape operations at manufacturing sites like refineries, power plants and steam crackers. In that survey, 83% of industrial IT and operations decision makers remarked how industrial AI has already played a major or significant role in their organization’s digital transformation strategy, with 79% adding that they already have an industrial AI project deployed. The results of a survey from Deloitte show that 93% of manufacturing companies likewise believe AI is pivotal to driving growth and innovation in the industrial sector.

The problem isn’t convincing decision makers to integrate industrial AI into their technology stacks or overall business strategy. The real issue — and what will determine how influential industrial AI remains in the coming months and years — is how to deploy it in a way that maximizes its value and ROI.

Identifying The Roadblocks And The Solutions To Overcome Them

Industrial AI isn’t a switch that suddenly opens a spigot of improved production processes, reduced energy waste or more value-added business outcomes. And it’s tempting to want to bolt industrial AI onto your existing operations and believe that’s all it takes. But getting the most out of industrial AI and capitalizing on its potential to truly transform the oil/gas and energy industries means we need to identify the biggest structural challenges that are inhibiting its value — and what it will take to overcome those roadblocks.

1. A Lack Of Data Quality And Management: One challenge stifling industrial AI’s maximum value is the quality and management of the data fed into AI models. Inconsistent data security formatting, large volumes of irrelevant data and disparate locations and silos for storing data all clog up an industrial organization’s ability to derive real value from data — or to ensure that the data is of high enough quality for AI.

AI models depend on quality data. If industrial operations are feeding low-quality or insecure data into these models, if the ability to access that data is hamstrung by its storage in disparate locations and if users don’t have the skills to adequately get useful takeaways from that data, then all of these problems will undercut the industrial AI application’s ability to deliver.

Data historians provide one solution to this problem. Using tags to both clearly identify particular data sets and format that data into standardized and secured formatting stages ensures a universal level of access to — and quality of — that data. Leveraging digital solutions like historians and tags to turn unstructured, unformatted data into secure, standardized, higher quality and easier-to-manage data means better quality data for the industrial AI model.

2. Technology And Team Silos That Stifle Collaboration And Visibility: Proper data quality and management and the ability to feed quality data into industrial AI models depend on widespread data visibility and collaboration between team members. But this visibility and collaboration remain elusive for many. Most AI experts and other data scientists and analysts working in the industrial sector largely operate in silos, with little if any collaboration between them.

This lack of collaboration has severe consequences. According to the results of our study, one of the most concerning is that industrial organizations have visibility into only two-thirds of their industrial data. Despite all the data collection tools and workflows that are aggregating huge quantities of industrial data every day, approximately one-third of that data is essentially invisible — all because of a lack of collaboration. That hurts industrial AI, too: When one-third of your industrial data picture is incomplete, the insights returned by the AI model will also be incomplete.

The solution to eliminating these team and technology silos and facilitating collaboration between teams starts with a top-down response from the C-suite to IT leadership. That vision has to call for new tools and workflows that don't just make it easier for teams to have visibility into each other’s data but actively encourage collaboration between teams. The more that everyone is on the same page, the better the industrial AI results.

3. A Lack Of Strategy Around Industrial AI Deployments: Deploying an industrial AI strategy can fuel a number of business benefits, including improved productivity, greater cost efficiency and reduced equipment downtime. And having such a strategy can prove to be a competitive advantage.

Not having an industrial AI strategy can result in productivity losses, greater downtime, cost inefficiencies and fortified data silos. With such possible downsides, the solution is clear: Business leaders need to plan their industrial AI deployments strategically, setting concrete ROI goals for specifically targeted industrial applications — even modest ones — that would most benefit from industrial AI.

Getting The Most Out Of Industrial AI

Industrial AI can unlock any number of game-changing benefits for the oil/gas and energy sectors: overhauling inefficient operations, increasing productivity and cost efficiencies, reducing downtime, curbing environmental and energy waste and presenting a possible competitive advantage. But this value can only be achieved by overcoming the major structural roadblocks that remain in the way. Refining data quality and management with next-generation data historians, eliminating internal technology and team silos and rolling out a strategy for how to best apply industrial AI to manufacturing processes and operations are the key steps needed for industrial AI to fully deliver its promised value.


Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?


Follow me on LinkedInCheck out my website