Artificial intelligence (AI) has been touted as both a potential savior and a potential destroyer of the world. Business leaders, however, are generally more concerned with their company’s survival than the fate of the world. And the type of AI they are pursuing is too narrow to end in world domination. Boston Consulting Group analysts Daniel Feldkamp and Joshua Kellar note, “There’s a reason AI has risen to the top of many executives’ agendas. It’s difficult to overstate the ways the technology has already changed the world — and this is just the beginning. AI is leading a global shift in human activity away from routine labor and toward creative endeavors, enhancing productivity, sharpening analytics capabilities, and vastly improving the consumer experience.”[1] At the same time, they admit, “Only 11% of organizations report significant financial value from AI.” What’s going on?

Avoiding Implementation Pitfalls

The amount of attention (or hype) received by artificial intelligence is understandable. AI was supposed to be the goose that laid golden eggs for businesses. In order to produce those golden eggs, the AI goose needed to ingest data. Journalist Thor Olavsrud (@ThorOlavsrud) explains, “In 2017, The Economist declared that data, rather than oil, had become the world’s most valuable resource. The refrain has been repeated ever since. Organizations across every industry have been and continue to invest heavily in data and analytics.”[2] In spite of all that investment, Jonathan Masci (@jonathanmasci), co-founder and director of Deep Learning at NNAISENSE, reports, “The results are in for artificial intelligence — and they aren’t good. The vast majority of all AI projects — 85%, according to a Gartner report — don’t deliver, or simply fall short altogether.”[3] Ouch! Masci quickly adds, however, “Yet for some, AI clearly works and adds value.”

While critics may want to claim AI is overhyped, Masci believes the problem more often than not is a business problem not a technological one. He explains, “Companies of all kinds — including manufacturers and industrial companies — are collecting huge amounts of data, and there’s no doubt that AI, if properly done, could help make production significantly cheaper, more efficient and safer. The question for companies, then, is how to reduce the risk of AI failure. And in many cases, it’s not a technology issue or a lack of good data; it’s a business issue. … Companies need to understand what could go wrong and what they need to do to avoid failure.” David Karandish (@dkarandish), Founder and CEO of Capacity, suggests four reasons AI projects go wrong.[4] They are:

1. Not enough data. Generally, finding enough of the right data is not a problem. Nevertheless, failing to ensure you have enough of the right data, and ensuring that the data is of the right quality, is a recipe for failure. Karandish notes, “AI is consistently learning and growing from its results and algorithms to provide better, more efficient, and more accurate outcomes in the future. For AI projects to learn, they need an abundance of information. The more data AI can ingest, the higher the accuracy of its output. Yet a common issue is a lack of sufficient data sets for developing AI solutions.”

2. Sticking with one means of learning. Humans use all of their senses to experience the world around them. In the same way, AI learns best when it experiences the world from a multitude of sources. Karandish explains, “You don’t just need a lot of data; you also need a lot of sources. For AI to work, it first has to learn. Limiting AI’s learning to one source or knowledge base can negatively affect how the end product operates. Without a range of information from different means, an AI solution will have gaps in its deliverables, causing issues for both creators and end-users.”

3. Lack of understanding from other employees. Businesses generally succeed when leaders focus on improving people, processes, and technology. Too often, leaders forget their people when implementing new technologies like AI. Karandish writes, “Not every person working on an AI-based project is an AI genius. However, successfully deploying an AI solution requires a general understanding by every employee and end-user. Everyone within an organization should understand the possibilities and limitations. With a lack of knowledge by all involved comes a lack of deployment.”

4. User experience is put on the back burner. Technologies are only useful if people use them. That truism shouldn’t have to be explained; however, Karandish notes that utility often relies on training. He writes, “If a user doesn’t understand the tool’s purpose or how to use it, they won’t buy it. Just as it is crucial to build an AI project that is accurate and efficient, you also need to make sure the interface is user-friendly. Customers and users need to realize the value of what you’re doing, how it solves the problems they are facing, and how they can use it with ease.” Feldkamp and Kellar add, “A critical element is people; value is unlocked through effective organizational learning and change management.”

Feldkamp and Kellar conclude, “Successful organizations deliberately reimagine the way they work with data and successfully move from pilots and proof of concept to scaling.”

Fostering Success

Tech journalist Maria Korolov (@MariaKorolov) reports, “AI adoption is on the rise. According to a recent McKinsey survey, 55% of companies use artificial intelligence in at least one function, and 27% attribute at least 5% of earnings before interest and taxes to AI, much of that in the form of cost savings.”[5] Regardless of what others are doing, a business should never adopt a new technology without making a business case for it. Masci notes, “To ensure AI success, [companies] need to understand what AI can actually do for them.” He also asserts that companies need to reel in their expectations. He explains, “Among the leading reasons for AI failure in manufacturing and industrial production are inflated expectations. Companies expect a sentient computerized being that can actually think, and are disappointed when they get nothing more than an enhanced version of the data analysis that they have been doing for years.”

According to Feldkamp and Kellar, the best way to control expectations is by defining what your company expects to achieve through its implementation. That means doing some homework. They write, “Harnessing the full power of AI is no easy task, but it’s an essential one for surviving and thriving now and in the years ahead.” They go on to suggest half a dozen actions companies can take to strengthen AI capabilities.

1. Define expectations. “Clearly articulate ambitions for value creation and desired outcomes while connecting AI to business strategy.”

2. Pursue digital transformation. “Treat AI as a business transformation, focusing not just on algorithms and tech but also on ways of working; design collaboratively with business owners using an agile approach (starting small, iterating, and evolving rapidly.” At Enterra Solutions®, we call this a “crawl, walk, run” approach.

3. Do your homework. “Learn from digital natives; the full power of AI comes from enterprise agility, bionic organization design, and continuously reimagined business processes”

4. Don’t forget about people. “Involve HR early on; consider AI needs in strategic workforce planning, launch flexible internal upskilling programs at all levels, rethink the value proposition to attract and retain employees, and diversify talent sourcing.”

5. Find the best data. “Decouple data-capability building from core IT transformations; embed flexibility by iterating and creating in-house capabilities.”

6. Protect your reputation. “Invest early in responsible AI governance, policies, expertise, and structural assessments; be transparent with stakeholders.”

Concluding Thoughts

Masci suggests using a little commonsense when embarking on your AI journey. He explains, “Like any other business endeavor, companies need to base their AI goals on their resources, including the data, funds and skills they can utilize. To do that, they need to develop an AI implementation plan, not just setting an overall goal, but determining what steps are needed to achieve that goal, avoiding problems that could stem from issues such as a lack of data, limited time frames for implementation, production schedules and more.” A good implementation plan ensures your company is on solid ground as it begins its digital transformation. Olavsrud concludes, “Insights gained from analytics and actions driven by machine learning algorithms can give organizations a competitive advantage, but mistakes can be costly in terms of reputation, revenue, or even lives.”

Footnotes
[1] Daniel Feldkamp and Joshua Kellar, “Artificial Intelligence: Ready to Ride the Wave?,” Boston Consulting Group, 2021.
[2] Thor Olavsrud, “6 famous analytics and AI disasters,” CIO, 23 November 2021.
[3] Jonathan Masci, “AI Has a Poor Track Record, Unless You Clearly Understand What You’re Going for,” IndustryWeek, 19 January 2022.
[4] David Karandish, “Why Artificial Intelligence (AI) pilot projects fail: 4 reasons,” The Enterprisers Project, 21 January 2022.
[5] Maria Korolov, “How to know when AI is the right solution,” CIO, 19 January 2022.