Despite mounting pressure on corporations to deploy some kind of AI strategy within the next year, a new survey conducted by Cisco finds that the vast majority of business leaders lack either the infrastructure, policies, talent, culture, or data to do it.

The survey, called Cisco’s AI Readiness Index, includes responses from more than 8,000 business leaders at companies with more than 500 employees across 30 markets. It finds that while 97% of respondents said the urgency to use AI technology at their companies has ramped up in the last six months—with more than half predicting negative consequences if they fail to act—just 14% are fully prepared to actually make that transition.

These findings reflect a significant gap between companies’ AI ambitions and the realities of how they operate. Not coincidentally, of course, Cisco is among the networking technology companies that stand to benefit from efforts to bridge that gap.

“AI has undoubtedly jolted the market and pushed companies to address and adapt,” Dev Stahlkopf, Cisco’s chief legal officer, tells Fast Company. “These findings are a timely, expansive analysis of where companies are making strides on AI readiness and where they still have a very long way to go.”

The survey assesses companies’ preparedness across 49 different metrics—from whether they’re ready for the increased power demands of AI training and tuning to whether their employees are receptive to the idea of using this technology. It finds that most companies are particularly unprepared when it comes to data readiness, with 57% of companies categorized as either having limited preparedness or being totally unprepared for the needs of AI.

One example of this lack of preparedness: Some 81% of respondents said their data is currently siloed within their company, inhibiting the extent to which it can be ingested by AI technology in the first place.

Another significant hurdle will be the governance of these systems, with just 3 in 10 respondents reporting that they have comprehensive AI policies and protocols in place. More than a quarter of business leaders surveyed, for example, say they have no systems in place to detect data biases, while another quarter of them say that even if biases are detected, they have no clear process to do anything about it. That’s despite the fact that a whopping 95% of respondents already have a highly-defined AI strategy in place. This suggests that while companies are racing ahead to develop new use cases for AI, they may not be adequately planning for the risks that come with it.

“Urgency to deploy AI-powered systems cannot be at the expense of effective governance,” says Stahlkopf. “The risks of deploying AI are real, but they are manageable when thoughtful governance practices are in place as enablers, not obstacles, to innovation.”

There are also more interpersonal challenges companies will have to overcome. Just 9% of the business leaders surveyed say their company’s corporate culture is fully prepared for the AI transition. Perhaps unsurprisingly, the survey finds that while 82% of respondents said their corporate boards and leadership teams are highly or moderately receptive to AI, just 68% said the same about employees. To address this obstacle, Cisco stresses the importance of training and putting people at the center of any company’s AI strategy. “When implementing AI at scale,” the report reads, “change management must have a people focus, ensuring open dialogue to address apprehensions and illustrate how AI can complement human roles rather than replace them.