North America dominated the AI market in 2019. According to Grandview Research, it accounted for over 42% share of global revenue, but when we talk about AI, we think of it developing in three waves. The first we call Artificial Narrow Intelligence (ANI), which is for discrete “one-trick pony” applications such as computer chess, predictive maintenance, or AI chatbots. The second, Artificial General Intelligence (AGI), is more nuanced and human, where AI provides more complex, multi-faceted decision-making. The third is what we call Artificial Super Intelligence (ASI), which is for when AI far surpasses human experts. We are somewhere between one and two at the moment, with multiple discrete systems starting to work in advanced ways with one another.
AI has reached a key juncture where the real-world benefits are instantly recognizable. One of the greatest advances has been the quality and interactability of data. In the industrial sector, AI applications are supported by a web of IoT sensors that create millions of meaningful data points to analyze, and in North America, 33% of financial services firms predict AI will change how they innovate, according to the Economist Intelligence Unit.
AI now enables high-value applications that humans would not be capable of delivering, such as predictive maintenance, using data to predict when a turbine may lose efficiency or, in one recent case, catastrophically fail – where predicting this saved a customer $34 million at a conservative estimate. More importantly, the North American AI market is expected to grow at a CAGR of 35.5% through 2025, according to OMR Global, because of significant investments in R&D and increased expenditure by the government and private organizations.
In terms of success, machine-learning AI is delivering significant results across a spectrum of industrial markets and asset types.
“The lift provided by AI for improved asset management is real,” says Mike Guilfoyle, vice president at ARC Advisory Group. “I’ve seen a multitude of wide-ranging results, from 50% maintenance work reduction and 5% increased reliability for elevators to $31.5 million in maintenance spend over three years for utility wind turbines.”
Fundamentally, companies adopt AI to become more efficient, reduce costs, and improve overall operations, including accuracy and safety. As machines and technology have progressed over time, the amount of data generated has increased exponentially. AI uses a wide variety of historical data to analyze trends, and this can help streamline and improve processes through cutting-edge solutions, AI-driven solutions, and operations scheduling. This, in turn, provides recommendations to people, thus substantially reducing errors and inefficiencies.
These various types of AI are applied in different ways throughout the industrial world to create targeted solutions provided as descriptive, predictive, prescriptive, and prognostic analytics. However, all of this requires human input to ensure consistent execution and accurate results. AI requires investment and commitment from companies to achieve the highest levels of success. If properly implemented and deployed, the payback can be vast.
A success metric commonly used with AI solutions is return on investment (ROI), where the value of the AI solution is quantified through a mathematical formula on a per-incident basis. For example, if AI provides early detection of an issue where downtime is avoided, then companies often calculate the value of the lost revenue from the downtime (had it occurred) as well as damage to equipment, other costs associated with contractual obligations, additional labor costs including overtime, and many other aspects.
Often businesses aren’t exactly sure what to expect from AI and sometimes over anticipate short-term success. Further, they may not follow best practices in implementing and operating their AI systems. Consequently, the combination of over expecting success and not following best practices can lead to underwhelming results.
To get the most value from AI, business leaders must understand what it can do today, as well as what it cannot provide. By matching business use cases to AI capabilities, companies can pragmatically leverage the power of AI in order to achieve maximum benefit with realistic and beneficial results. In North America, for example, AI is being more actively applied by customer services, R&D, and manufacturing and operations, according to MIT Technology Review.
In order to get the most value from AI, companies need three primary things: 1) a corporate culture that encourages and facilitates AI infusion into business processes, 2) an understanding and trust of the power of AI technology, and 3) an IT infrastructure that provides the underlying data requirements and processing power needed by AI. Without all three, barriers will inherently exist to achieving maximum success.
AI is rapidly evolving from providing limited applications in specific industries and in very targeted areas to becoming a broader “intellect” that can transform businesses through integrated AI solutions to solve more and more complex problems. The pervasiveness of AI is a growing trend and one that will endure for the long-term future as AI evolves and improves in its applicability to real-world challenges.
As AI becomes more powerful and more prolific in business, more and more tasks are suited to its benefits. AI has often been associated with process analytics where it provides early detection of issues and offers guidance to resolve them quickly. Many of the major advancements have been in this area including predictive operations, prescriptive analytics, and prognostic (forecast driven) maintenance. However, AI is now becoming pervasive in all areas of the industrial world, from engineering design to production operations to automated maintenance to performance management.
The amount of data required by AI varies based on the type of solution being provided. Some types of AI need a lot of data, others not as much. If the AI is analyzing a wide range of changing processes or is impacted by the four seasons, then significant data may be required, such as one year’s worth of time-series sensor data. However, in more controlled situations, much less data is needed to define the scope of what is being analyzed. As AI technologies evolve and combine, less and less data will be required to get started with AI.
About the Author
With over 30 years of experience in the industrial software sector, Jim Chappell is currently head of AI and Advanced Analytics across all AVEVA business units, products, and markets. Prior to his current position, he led the Asset Performance Management (APM) suite of software products and related engineering/analytics services for Schneider Electric.
AVEVA is a global leader in engineering and industrial software driving digital transformation across the entire asset and operations life cycle of capital-intensive industries. The company’s engineering, planning and operations, asset performance, and monitoring and control solutions deliver proven results to over 16,000 customers across the globe. Its customers are supported by the largest industrial software ecosystem, including 4,200 partners and 5,700 certified developers. AVEVA is headquartered in Cambridge, UK, with over 4,400 employees at 80 locations in over 40 countries. For more details visit: www.aveva.com
Copyright © 2021 IDG Communications, Inc.