As organisations move from experimenting with artificial intelligence to embedding it in business operations, the question is increasingly shifting from whether they should use AI to whether they can trust the systems they are putting into production.New research from SAS, with insights from IDC, suggests that the answer could have a significant bearing on the financial returns organisations achieve from their AI investments. The second annual Data and AI Impact Report: The New Economics of Trust found that organisations applying trustworthy AI practices were 15 times more likely to report strong or high returns on investment from their AI projects.The finding points to a broader issue facing technology leaders. AI adoption is no longer simply about acquiring models or deploying new applications. Organisations also need the governance, data quality, oversight and accountability structures required to ensure that those systems can be relied upon.The report, based on a survey of 2,699 decision-makers across 28 countries and four industries, found a substantial gap between organisations that have established trustworthy AI practices and those that have not. Organisations investing in trustworthy AI measures were 15 times more likely to report strong or high ROI, at 62% compared with 4% among those with weaker practices. That difference suggests that trust is not necessarily a constraint on AI adoption. Instead, the research indicates that the processes organisations put around AI can influence how effectively they turn deployments into business value.“When AI works, it’s incredibly impactful,” said Bryan Harris, CTO at SAS. “However, it is well documented that state-of-the-art agents can have error rates that exceed 25% on complex tasks – which is unacceptable in high-stakes decision-making. In order to achieve accuracy and repeatability, organisations must embed domain expertise into agentic workflows, while keeping people at the centre of governance and oversight. Organisations that do this successfully will close the trust gap and gain a competitive advantage in the market with AI.”The challenge becomes more pronounced as organisations move towards agentic AI, where systems can perform increasingly complex tasks with less direct human intervention. According to the report, trust in AI falls from 76% for generative AI to 66% for agentic AI. At the same time, 97.2% of users globally override AI-generated recommendations in at least some cases. The leading reason for those overrides is a lack of explanation for how the AI arrived at its decision.This creates a practical problem for enterprises. An AI system may produce an apparently useful recommendation, but if employees cannot understand why the recommendation was made, they may be reluctant to act on it. Repeatedly overriding AI also undermines one of the principal reasons organisations deploy the technology in the first place: improving productivity and decision-making at scale. The report notes that manual corrections resulting from a lack of trust can consume time, reduce productivity and affect profitability.For technology leaders, explainability therefore becomes more than a technical feature. It becomes part of the operational relationship between employees and AI systems.“As AI becomes more autonomous, organisations face a new challenge: maintaining confidence in systems people don’t fully understand,” said Chris Marshall, Vice President at IDC. “Our findings show that stronger oversight, explainability, accountability and data foundations are becoming prerequisites for scaling AI successfully.”The report also highlights a less visible part of the AI trust problem: the quality and maturity of the data infrastructure supporting these systems. Only 17.5% of enterprises surveyed have fully optimised data infrastructure considered mature enough for the demands of agentic AI. Organisations with an optimised data foundation were four times more likely to expect strong ROI from AI projects and six times more likely to mandate the data quality and explainability controls required to build trust.This reinforces a familiar challenge for CIOs. AI systems can only operate within the constraints of the data, processes and governance surrounding them. Deploying increasingly sophisticated models on fragmented, outdated or poorly governed data can therefore create a mismatch between the sophistication of the technology and the reliability of the decisions it produces.Trustworthy AI consequently begins well before a model reaches production. It involves establishing confidence in the underlying data, defining who is accountable for AI-driven decisions and putting controls in place to monitor how systems behave.The report defines trustworthy AI across five dimensions: data quality and governance; model governance and oversight; explainability and fairness; responsible AI policy; and audit and accountability. Organisations classified as trustworthy AI leaders achieved an average score of at least 80 out of 100 across these dimensions.
This page contains an attributed headline and the preview text supplied through the publisher’s RSS feed. Copyright in the original reporting belongs to CIO Africa.
