Dive Brief:
- More than half of global enterprise tech leaders say the CIO is held responsible for AI errors within their organizations, according to a Tuesday report from 8x8 Inc. The business communications firm surveyed more than 2,500 tech executives in the U.S., U.K., France, Australia and Ireland in July for the results.
- CIOs have risen to the top of the blame list for AI outputs, more so than other department leaders, such as 16% of customer service and 6% of legal leadership who say they are on the hook when AI fails, 8x8’s survey found.
- Fragmented governance and documentation processes can contribute to a lack of control that CIOs have over their company’s AI uses, Samuel Wilson, 8x8 CEO, told CIO Dive in an email. "There needs to be more structural guarantees that the CIO actually sees what the AI did before, during or after an incident,” Wilson said.
Dive Insight:
Despite a lack of trust in AI systems, most enterprises are forging ahead, embedding them in core business processes across the enterprise, even if they don’t have clarity on who owns the technology.
The pressure to successfully adopt AI is weighing heavily on tech leaders — 61% said they fear losing their job if they fail to lead their organization through the AI transition, according to an April Writer survey.
AI ownership has been a pain point for organizations as long as they’ve been deploying the technology. Accountability for its success most often lies with CIOs, CTOs or other tech leaders, an April Altimetrik report found.
Although CIOs are being held responsible, they often lack visibility into all of the tools and processes that their company is using — and rely on others to keep them in the loop, Wilson said. They also are often not in charge of choosing vendors and setting up an audit trail.
Many companies look at accountability the wrong way, he said.
“They check whether the AI got the right answer, but nobody checks why it got there,” Wilson said. “A CIO who only sees outcomes is reading the last page of a book with no chapters. You can't defend a decision you can't reconstruct.”
If CIOs are going to be held accountable, they can take steps to gain more authority over their AI processes. Leaders should work with their IT teams to include continuous IT and security reviews of AI vendors, Wilson said.
"CIOs should also have their teams get tougher with vendors,” he said. “If a vendor can't show what their model is doing in real time, they're asking you to sign away accountability without giving you the tools to keep it.”
Tech leaders can also work to form a committee with leaders from IT, legal, compliance and the business units that deploy the technology, so accountability is settled before something goes wrong, Wilson added. Quality management has been a part of business practices for decades, and a similar process can be applied to AI agents, he said.
