All the talk about AI and everything it can do for businesses may make it feel like most companies are already using the technology, but a new study from K2view shows that isn’t necessarily the case. The data preparation company found that as of July, just 2% of enterprises in the U.S. and U.K. actually have deployed AI solutions. What’s the holdup for everyone else? Other than cost, it’s issues dealing with data.

The problems that enterprises have with their data are myriad. Nearly half say they are concerned with data security and privacy in a generative AI system. And a third of all companies are still working on data readiness—a top issue for 45% of all companies with 5,000 or more employees. The data issues also cover many areas. A total of 48% of companies see problems with scalability and performance. Issues with quality and consistency, as well as real-time data integration and access challenges are major difficulties for 46% of respondents.

Companies identified many valuable internal data sources for potential customer-facing AI integrations, meaning there are a variety of different databases and data types to sync up. Customer interactions—which can run the gamut from invoices to emails to chats with a helpdesk—are valuable to 56% of companies. But almost as many see CRM customer data and marketing data as key. Financial data, which is likely to be more standardized than some of these other types of information, is in a distant fourth place, with 36% of users seeing it as a top source for AI.

These results illustrate a common issue many companies are trying to deal with when transitioning to AI. Data is only useful when it can be accessed, its information recognized and “understood” by the AI system, and the information can be used to determine a customized outcome. Though most businesses have data dating back to when they first started using computers, it can take a lot of effort to make it useful. But while data prep work may end up being just as taxing as programming generative AI applications, studies have shown that the efficiency of an AI system may be worth the effort.

It’s important to note that generative AI isn’t the only way for companies to become more efficient—and it might not be the best solution. I talked to Nick Elprin, CEO of AI operations platform Domino Data Lab, about how to determine whether an older version of data science can solve your company’s issues better. An excerpt from our conversation is later in this newsletter.

Nominations for Forbes’ CIO Next List are now open. If you know any CIOs who are making an outsized impact on their company or business, please let us know about them here by November 8.

Jakub Porzycki/NurPhoto via Getty Images

Hackers have found a quiet way to deploy malicious code to steal information and poison data: Hide it in files uploaded to........

© Forbes