Quarter of Execs Say AI Errors Exposed
A new survey by Workiva found that 84% of executives trust AI results without human review to some extent, but only 11% believe their data is of high enough quality to use this technology.
The survey also revealed that one in four executives said internal audits detected AI errors that reached the public or board members.
According to the report, this discrepancy between executives’ perception and actual practice is a concern, as AI transforms businesses at an unprecedented pace, introducing a new risk: the pace of this transformation may outstrip leaders’ ability to oversee it.
Quality of Data Threatens AI Accuracy and Investor Trust
Executives largely agree that poor data quality limits the use of AI in their organizations, suggesting that widespread adoption of AI depends on having reliable and controllable information.
Institutional investors share these concerns and also question the accuracy of AI-generated content, with only 11% of executives considering their data to be of high enough quality to use AI.
27% said poor data quality significantly hindered AI implementation in key processes, and 71% reported that poor data quality at least moderately affected the use of AI for financial and sustainability reporting.
Barbara Larson, CFO of Workiva, stated, “Trusting AI without controlling data quality is a risk, not a strategy, as CFOs need platforms that connect AI with reliable and auditable information, so they can verify every result and support every data point they present.”
Related: Alipay+ expands with new Asia Pacific bank partners
Executives Demand Tools to Monitor AI
As AI use grows, executives are calling for tools to oversee it, with financial information teams, and especially CFOs, needing tools to control AI results as much as the technology itself.
They foresee a greater need for specific infrastructure and highlight several priorities, including systems that maintain official accounting records, such as general ledgers, which 49% will continue to need.
45% will need solutions that enable information tracking and facilitate audits, and 55% will need platforms to manage automated agents and processes.
Jason Darby, CFO of Amalgamated Bank, stated, “Generic AI is not enough for financial information, as investors, regulators, and boards expect reliable answers, and the real advantage lies in using AI developed for these specific needs, based on controlled and auditable information, and complemented with human judgment.”
This statement highlights the importance of reliable and auditable information in AI systems, which is a key concern for executives and investors alike, as they seek to verify AI results and support the decisions and information they present based on them.
The survey’s findings suggest that companies that do not address data quality concerns and investor expectations risk falling behind more proactive competitors.
As the use of AI grows, they will need to prioritize data quality and invest in tools that enable them to monitor and control AI results, or risk being left behind.
