Three thousand spectators line the Portsmouth waterways to watch the launch of the world’s first globe-circling yacht race. Among the nineteen ships in this inaugural 1973 Whitbread Round the World Yacht Race, now known simply as The Ocean Race, was the Sayula II and its skipper, Ramon Carlin. The owner of a popular dishwashing machine manufacturing company in his home country of Mexico, Ramon only started recreational sailing two years prior.
When registering for the race, he was asked what ship he’d be entering. Not even owning a boat, Ramon’s answer was, “I don’t know, but you can be sure I’ll be there when the regatta starts.” Not only did Ramon not yet have a boat, but he also didn’t have a crew to help him on the four-leg, nine-month journey. He turned to family, friends, coworkers, and a diverse assortment of strangers. The plan was for his motley crew to learn and adapt between the launching point of Portsmouth and the end of the first leg in Cape Town. The competition quickly dismissed Ramon, the Sayula II (Ramon’s recently purchased ship) and its crew. These predominantly European yachtsmen were confident the key to winning this new type of race would be their undeniable mastery of traditional sailing methods.
What separated Ramon and his crew from the eighteen other entrants was not a bigger or faster ship. It was his strategy and execution – knowing this race was unlike anything before it. Simply emulating the traditional style of his peers or being myopically focused on techniques that contributed to past success wouldn’t win the race. Ramon maneuvered outside of the traditional shipping lanes followed by his peers. His leadership diverged from his fellow captains as well; both in fostering a positive culture and by balancing the focus of his crew – each member was able to give attention to several aspects of the boat’s operations.
When looking at the use of artificial intelligence and machine learning, the need for companies to apply both to their business solutions to remain ahead, and often just at pace, of their competition grows by the day. Yet there is looming intimidation and risk when you hear Gartner’s famous claim of an 85% failure rate of AI/ML business solutions or a lower, nevertheless staggering, IDC rate of 50%.1,2 But why such a high failure rate when there is so much investment and effort across a range of industries and players? What are the ingredients of a successful AI/ML business solution?
The primary reason models fail isn’t because the science is unsound or the statistics are unable to render meaning. The primary reason for failure starts at the AI/ML solution’s purpose. A common misconception is the purpose should be the identification of statistical signals within data and the subsequent use of those signals to predict a future outcome: a solution narrowly focused on the science alone. Comparatively, successful AI/ML solutions start at a business outcome such as increased lead conversion, narrowed forecast accuracy, or reduced customer churn which are then addressed with workflows or prescribed actions that are supported or driven by AI/ML. The focus should start and end with the business. Ultimately, the accuracy of a model feels important to success (and is) but is second to how that model is presented to the user, integrates into business process, and ingests and reacts to feedback.
The cause for this misdirected focus is the predilection to frame new concepts and solutions in a context with which we are familiar. AI/ML in business is often sold and implemented as a new tool or feature to productionalize; no different than moving to the cloud to save overhead and reduce downtime or implementing new services to increase productivity. Therein lies a course plotted toward failure. Successful AI and ML solutions are more than just science, math, and technology. They are, first and foremost, a transformation of business processes.
During the Whitbread Round the World Yacht Race, the onboard radio of the Sayula II malfunctioned during the last leg. Ramon and his crew had no knowledge of their race position as they made their way toward Portsmouth. Nine months after their 27,000 nautical mile journey began, they sailed into port accompanied by a small fleet of spectator boats and a handful of helicopters. They arrived to a cheering crowd of four thousand as the winners of this historic first race.
The race for AI/ML adoption in business has striking parallels to this round-the-world yachting competition. Imagine those final weeks for the Sayula II crew. No radio. Months of Herculean effort and tens of thousands of miles invested. They could look out to the ocean all around them without a chance of locating their competition’s position. They were guided by the faith that their strategies would pay off. They would only find out if their faith was well-placed once the race was over; when there would be no more opportunity for adaptation and execution.
All industries are in a race to adopt AI/ML solutions in order to adapt to new business paradigms — a race unlike those that have come before. Is your organization navigating the uncharted seas of AI/ML based on principles of the status quo or is it part of Sayula II’s innovative course to victory?
1. Gartner Says Nearly Half of CIOs Are Planning to Deploy Artificial Intelligence
2. IDC Survey Finds Artificial Intelligence to be a Priority for Organizations But Few Have Implemented an Enterprise-Wide Strategy