Do you believe these five myths about scaling AI?

You're already experimenting with AI in your company. Perhaps you've embarked on a pilot program that applies technologies and techniques such as machine learning, natural language processing and computational intelligence to solve a specific business challenge. You're testing a capability with real users and data, intent on paving your company's path to future growth. Over time, if the pilot yields solid results, you'll want to continue the initiative.

The key is to determine what's next—how to expand the value—and that requires successfully scaling the AI effort.

<<< Start >>>

<<< End >>>

What does it mean to scale AI? Basically, it's extending an AI capability from an initial pilot program to the widest strategic scope and impact, bringing the most value to an organization.

But at too many companies, AI initiatives hit major roadblocks after the proof of concept, even as executives recognize that scaling AI is a major priority.

Fortunately, recognizing and sidestepping AI myths can help leaders get on the right track. Check out five common myths—and realities—that every would-be AI adopter should keep in mind:

Myth: Scaling should happen as quickly as possible.

Effectively scaling AI typically doesn't happen in a matter of hours, days, weeks, or even months. Doing it right doesn't mean doing it fast. Successfully scaling AI means doing it purposefully... and that can take years.

We conducted a global study—surveying 1,500 executives across 16 industries—and found that companies that have had the most success scaling their AI efforts are 65 percent more likely than their less successful peers to set timelines of one to two years.

Some take even longer. Take, for instance, the case of a Japanese communications service provider. With our help, the company took a measured yet aggressive approach to implementing AI-powered analytics in every aspect of their business: targeting of customers, retention of existing customers, and determining where to build out their next-generation mobile network. In the third year of their journey, they are achieving tremendous business results.

You could say this company and others like it demonstrate "sticktoitiveness": they commit to a goal and don't waver when results aren't immediate. Ultimately, their patience and measured persistence are rewarded.

Myth: Spending more improves the likelihood of success.

We've found that companies that scale AI successfully actually spend less than their less successful counterparts. This is in part because, again, they're more careful and thoughtful in their approach to AI. They rely on clearly defined strategies and operating models to make enterprise-wide AI a reality.

But scaling AI effectively costs less than you might think because the technology itself is already being delivered at massive scale today by major technology providers. For example, chatbots are fairly inexpensive to implement yet can result in major ROI.

In fact, we recently helped two businesses, both mobile network companies, introduce virtual assistant chatbots into their customer service operations. Those bots now handle millions of customer calls each year, saving our clients tens of millions of dollars in call center costs.

Myth: The more data, the better.

It's not unusual for company leaders to share concerns with us that they don't have the right data, or they don't have enough data for successful scaling. Of course, companies aren't wrong to emphasize the role of data in implementing AI technologies like machine learning and robotic process automation.

<<< Start >>>

But we find that they often get off track when they focus on data issues before determining whether they have the right business processes in place. When companies ensure the latter, it becomes easier to approach data issues strategically and to identify and structure the information necessary for AI initiatives.

We recently helped a tech client do just that with their sales operations. The client sought to automate their commission payments to sales executives. Before working with them on a solution, we first examined how they tracked their sales data, how they targeted clients and other business practices. What we found were inefficiencies and broken processes. With our guidance, the client fixed and streamlined those processes.

It was at that point that we could evaluate what data the client truly needed... and we determined that they already possessed the information necessary for fine-tuning their commission process. Armed with this data, we devised a solution that ultimately saved them hundreds of millions of dollars while also leading to improved sales.

Myth: AI operating models should solely reside in the IT department.

Too often, we meet with company leaders who believe an AI initiative should be launched by a chief technology officer or a chief information officer. But the companies that scale AI successfully have CEOs who actively champion the technology themselves, focus on a specific business outcome, and make clear that AI is a priority for the entire organization.

What's more, successful AI efforts are often undertaken by an interdisciplinary team. They involve not just engineers and IT professionals, but also data scientists and business experts to ensure that the technical architecture of the initiatives meets the company's needs. And as AI is adopted, members of just about every business function, from sales to supply chain, are included to help the implementation roll out across the organization.

Education across an enterprise can go a long way toward encouraging a multi-disciplinary approach. We worked with one fast-food chain, for instance, that had planned to adopt AI in its IT operations. We showed the company that AI had wider applications and invited leaders from various departments to innovation workshops so they could understand the true power of AI. The more they learned, the more enthusiastic they were about scaling the technology. Today, the company uses AI to determine how, where and when they deliver food, transforming their business model.

Myth: AI scaling is harder for larger organizations than smaller ones.

Conventional wisdom dictates that smaller companies are always more agile and therefore are in a better position to take a new technology and run with it. Larger companies, in contrast, have many more moving parts, which may lend itself to the perception that they struggle to adopt change.

Yet when it comes to scaling AI, giant entities are, in fact, no less successful: When we grouped companies in our study by size, we found no significant relationship between scaling success rates or return on investment.

Why is this the case? It's important to remember that large companies have grown to their current size for a reason: They know how to scale, and they're good at it. They apply those capabilities to their AI initiatives. And like any other company that journeys beyond proof of concept, they're using the proven strategies to do it. When it comes to AI, they've separated fact from fiction, yielding very real results.