This post was co-authored with David Immerman, Senior Research Analyst, and Chris Joynt, AI & Analytics Translator
Artificial Intelligence (AI) continues to garner attention in mainstream media and pop culture. Increasingly consumers interact with the technology in their personal lives and are inspired by its potential. While consumers are awestruck by a new voice recognition software on the latest smart home device, businesses are struggling to leverage AI capabilities that generate meaningful value.
Three out of four organizations report that "business adoption" of AI initiatives is a significant challenge.
But why?
In many cases, projects stall or fail because they are not tied to business goals while, at the same time, drain valuable resources - a phenomenon known as pilot purgatory. Additionally, many fall victim to the temptation of viewing artificial intelligence in business as the newest panacea for all business problems; a distinct end state that can be achieved upon the discovery of the perfect algorithm.
Instead, AI should be viewed as an enabling capability in a more comprehensive digital transformation (DX) program.
A better understanding of the uses, constraints, and potential of AI can lead to improved resourcing and alignment with DX goals, improving the likelihood of project success and ROI.
There are three high-level steps companies should undergo when evaluating AI’s potential role in their organization.
Read below for more specific guidance on these foundational elements to implementing AI in our organization.
Business pain points typically are key value areas inhibiting companies from realizing the entirety of their strategic goals. For manufacturers it could be sub-optimal levels of overall equipment effectiveness (OEE) impacting production efficiencies, excessive amounts of defects dampening product quality and increasing scrap, or frequent and expensive asset downtime resulting in excessive service costs and lower customer satisfaction. Even minor improvements in some of these metrics can drive massive gains across the value chain.
Identifying which problem is most pressing and has the greatest impact on financial and operational metrics is a solid starting place to align a digital transformation strategy to. Narrow the scope to these pressing areas and identify any high-level themes that are consistent with those opportunities. This can form the basis of a comprehensive AI strategy to be developed in the future.
A well-designed AI program is tightly aligned to an overarching DX strategy and seeks to deliver on multiple discrete AI use cases. In this sense, AI is a means to an end. Whether it’s incorporating an AI-driven self-service capability in a new digitally-enabled product, using predictive capabilities to transform field service, or even optimizing internal processes like supply chain management with AI -- all are consistent with an overarching DX strategy of being more agile and responsive to customer needs. Put another way, none of these are “science projects”.
We’ve observed that those DX programs that take a financial-impact first approach with use cases tied to business strategies and goals tend to drive the greatest value in the shortest time. Many companies attempt to simultaneously roll out dozens of use cases and quickly run in to pilot purgatory.
For artificial intelligence in business, the concept is similar - business leaders may be overjoyed at the universe of opportunities to exploit their data, which can lead to scope creep and AI projects that seek to satisfy lofty ambitions right out of the gate. The chance of success is exponentially higher for companies with a financial-first approach, rolling-out one or a few prioritized AI use cases for a pressing problem or high-value area rather than applying AI to every piece of accessible enterprise data. Establishing this initial scope is key to fostering a successful AI program with longevity.
The scope of that AI program should be in service of a digital transformation goal. Business leaders should stop and ask, how does this AI project give my organization a new capability that will help us achieve DX goals?
Once you have identified the pain points and aligned your AI program to an overarching DX strategy, it is time to develop an AI use case. An AI use case should have a few key elements.
First, it should capitalize on data. DX programs by nature create and source lots of data with the potential to be exploited by AI for efficiencies. AI use cases leverage this data to get maximum value from it.
Second, it should deliver timely actionable insight, tailored to a user experience. If AI is going to have any real-world value, it must improve decision making (whether that decision is made by a human or a smart system). Trends, anomalies predictions, recommendations, or any number of outputs from an AI system usually pertain to a specific point in time. It’s imperative to deliver this insight to the right person or system at the right time before it becomes useless. It also should be delivered in the context that is most helpful for a user or system trying to make a better decision. UX design thinking can help to make sure that insight achieves maximum impact.
Third, it should collect more data to measure outcomes and track improvement. We know AI applications thrive on data. But they also present an excellent opportunity to structure the data we are collecting, and begin an upward spiral of AI development to scale from.
For example, an AI application that can predict equipment failure is valuable. However, an AI application that can predict equipment failure, and incorporates feedback the occurrence of failure will learn and improve it’s predictions is immensely more valuable. An AI application that predicts failure, measures actions taken against that prediction, and the ensuing outcomes of those measures will not only learn and get more accurate over time, it will also structure the data to provide a pathway to make better business decisions for automation and optimization.
With a strong AI foundation, industrial companies are well-equipped to make data-driven decisions and solve problems across the value chain. Leveraging AI for these common daily occurrences will drive the most meaningful impact for industrial companies and expedite adoption internally.
Artificial intelligence in business will reach its transformative potential when the data-driven insights they inform are democratized across the enterprise. Industrial companies following these key considerations can skirt AI pitfalls and drive the most attainable business value in shortest time frame.