GenAI Benefits Minimal Now but Should Accelerate

Though the initial benefits from generative artificial intelligence have been marginal for most firms, the technology is on the verge of accelerating functional effectiveness and competitive differentiation, according to a new report from the CMO Council.

In fact, 79 percent of senior executives believe genAI will deliver a competitive advantage within 18 months. And, for those who are prepared, the advantage is clear: 81 percent of confident leaders expect genAI to improve customer experience and unlock new growth.

But at the same time, 30 percent of genAI projects are expected to fail post–proof of concept due to many long-standing barriers.

The study found that the potential for increased business value with genAI will challenge even the most seasoned business leaders because the underlying data isn’t prepared to work with the technology.

The study, a collaboration between the CMO Council, the Business Performance Information Network, the Growth Officer Council, and Encompaas, also shows that three-fifths of business leaders lack confidence in their data-AI readiness to achieve genAI business value. Only 13 percent are extremely confident in their data-AI readiness.

Most confident business leaders come from companies that have made a data transformation that supports genAI, according to the study, citing an unrelated Gartner study:

“Data management skills and processes, as well as underlying technology and architecture, must undergo an evolutionary change to properly enable AI initiatives,” Gartner says.

To harness genAI’s capabilities, organizations first have to discover, classify, manage, and secure data for genAI to query, the study says. “Many organizations need to shore up their data environment amid an unprecedented data deluge and growing security and privacy concerns.”

Though today most firms are comfortable enough with their structured data to run SQL queries, genAI queries are much more complex because they use unstructured data, says David Gould, Encompaas’ chief customer officer.

For example, executives might want to use genAI to examine business contracts, requests for proposals, accounts payable documents, travel documents, and similar content, Gould explains. “That’s a whole lot of different requirements that are coming into play there. There’s a lot more complexity to it. You need to consider if you’re using the right data to produce the outcome that is desired.”

Gould, citing Gartner research, says that 60 percent of genAI projects fail once they go into production because of poor data quality and siloed systems; inaccurate or incomplete data fueling AI models; security, compliance, and governance blind spots; and a lack of explainability, transparency, and trust in AI outputs.

The research adds that it’s critical for organizations to ensure data meets specific requirements, including representativeness, responsiveness, and proper data governance principles.

In a business contract, for example, important data will be on several different pages. The genAI needs to be trained to find the relevant data wherever it lies.

“Companies need a state of data-AI readiness that delivers strong performance in data quality, accuracy, reliability, security, and privacy,” the study says. “Far too many companies lack confidence in their ability to get to this state, so we can expect a high failure rate.”

Gould adds: “There’s a general understanding now, particularly with large enterprises, that data quality or data readiness is a key factor in making generative AI work.”

In addition to the business contract example above, many companies use collaborative programs such as SharePoint that has data that needs to be vetted, leading to other complications, including potential hallucinations, according to Gould.

According to the CMO Council, to climb the data-AI maturity curve, companies will need to do the following:

• Invest in a modern technology stack and data supply chain to support ambitious genAI projects.
• Create a culture of AI and data literacy to figure out and be aware of what data points are important to genAI.
• Make innovative yet sensible decisions on use cases where genAI can provide the best value.
• Build a track record of success that engenders trust for genAI among employees, customers and partners.