Here’s the real reason 75% of enterprise AI initiatives fail

Since launching Generative AI, we’ve conducted extensive research involving CEO interviews and deep dives into leading companies. This work provided insider knowledge of the success of AI initiatives and culminated in the book Human machines. Here are some of those perspectives.

One AI does not fit all

Artificial intelligence can be copied – and one size does not fit all. What is not copyable is a unique business model, processes and integration of people with that technology.

Our research finds that the massive rush to apply AI technologies to existing business models and legacy processes will not lead to success.

Spencer Fung, CEO of global supply chain giant Li & Fung, offers an analogy: “Companies acquiring AI without a new business model are like a company digitizing a horse and carriage – while the competition has created a digital automobile “.

Adding AI to a business model of the past does not lead to competitiveness, it simply solidifies old processes. AI is essential but insufficient to provide a competitive advantage. Before attempting to integrate AI into their businesses, corporate leaders need to reevaluate and update their business models.

Data doesn’t hold up against volatility

AI relies on historical data that may not be reliable in unpredictable and ever-changing global business environments.

“Every model based on mathematics collapsed when the pandemic hit. None of the assumed parameters can be trusted,” as John Sicard, CEO of supply chain software leader Kinaxis, told us.

Business decisions are not made in a vacuum separate from labor, inflation and geopolitical issues. Experienced workers bring domain expertise and deep knowledge of their environment. They step in when digital analytics aren’t enough like a pilot who takes control in unusual circumstances.

This knowledge is essential – and ignoring its value is fraught with danger. Sicard sums it up with this warning: “Blind obedience to the model is dead. It pulled us off a cliff during the pandemic. It’s reckless.”

This echoes our recent discussion with chess grandmaster Garry Kasparov, the first chess player defeated by a computer. Although Kasparov concludes that machines are better than humans 95% of the time, humans need to know When and How to intervene the remaining 5% of the time. This is critical.

Kasparov notes that the advantage comes to the person who knows when to rely on instinct and intuition. This is the difference between a good decision maker and a great one. “A little change here and there has the biggest payoff. We should not dispute the superiority of the car in 95% of cases. But we do in the other 5%,” he explained.

It’s also important to know when to be humble enough to let the algorithms work autonomously. AI tools don’t have the ability to understand context, but they shouldn’t.

This insight helps leaders understand the critical human elements that drive successful AI implementations. As Ted English, former CEO of TJX Companies, a Fortune 100 apparel and home fashion retailer, says, leadership requires “a lot of instinct, experience and knowledge. You can’t get that from a car.”

AI requires new human skills

As AI becomes commonplace, companies must cultivate new human skills in their workforce. In our interviews with executives, we heard repeatedly that the new competitive advantage comes down to “human interpersonal skills,” “human creativity,” and “personal relationships.”

Peter Cameron, CEO of Lenox, told us: “Nothing replaces long-term relationships that are personal – and the longer the relationship, the better.”

Rod Harl, CEO of Alene Candles, a company with 80% revenue growth over five years, said their best decision was to invest in training employees in interpersonal skills and mindfulness techniques. Combining these skills with human creativity, Harl notes, “is the secret sauce.”

As Maria Villablanca, co-founder and CEO of the Future Insight Network, said: “Companies need people who can be creative and innovative in how they find solutions. Companies need creative problem solvers with interpersonal skills. Cars can’t compete with that.”

As AI takes on more tasks, there is a risk of skill atrophy and knowledge loss. In addition to retaining experienced talent, companies must consider ways to develop decision-making skills within human resources.

Today, the human skills considered most important by leaders are interpersonal skills: basic conflict resolution, communication, emotional detachment, and mindfulness practices. While digital literacy is expected, effective interpersonal skills are a priority. These uniquely human abilities are in short supply and may require training.

More mandatory comments posted by wealth:

Opinions expressed in Fortune.com comments are solely the opinions of their authors and do not necessarily reflect the opinions and beliefs of wealth.