Widespread Experimentation Has Yet to Produce Enterprise-Scale Adoption
The pressure is on for enterprises to get artificial intelligence systems into production, but that doesn’t mean business decision-makers trust the information that AI provides.
See Also: Rise of Malicious AI Skills Expands Enterprise Risk
Nearly all of the data and analytics leaders who participated in a recent survey on how data teams are using AI said that their companies are using or exploring AI for analytics, and 56% of respondents have put the technology into production. However, only 7% said that AI analytics are actively used in every line of business across the enterprise.
“Enterprises have spent years building trusted dashboards and processes around their data, so they are not going to replace them with AI simply because the technology is available,” said Soham Mazumdar, CEO and co-founder of WisdomAI, in a statement.
WisdomAI surveyed 201 senior AI, analytics and data leaders at North American companies with at least $1 billion in annual revenue. Respondents included chief data officers, chief analytics officers and executives responsible for business intelligence, data engineering and AI transformation.
Confidence in AI-generated analysis remains uneven. Only 19% of respondents said they are very confident in the results produced by their systems. Another 46% are somewhat confident, but 35% said they were not very confident or had no confidence at all.
With little trust in AI systems, many respondents said employees still rely on the technology tools they were using before AI entered the picture, with 81% saying they still rely on dashboards or one-off requests as their primary method of data delivery. Conversational business intelligence and agentic workflows were the primary method for only 19% of respondents.
Among the 113 organizations that reported having AI analytics systems in production, 54% said they’re using AI-automated data visualization and 43% are using natural language summaries. Multi-step reasoning, proactive agents, and insight-to-action workflow automation are rare, the survey found.
Speed and accessibility were the greatest operational benefits AI brought to data teams, the survey found. AI increased access to insights for 56% of respondents and half said it increased insight accessibility for non-technical users. Improved accuracy and reliability was cited by 26% of respondents.
At the same time, 53% of respondents said it took more than a day to fulfill an analytics or dashboard request, and 26% said it would happen within a day. Only 8% said answers were instantaneous.
When asked what would improve the accuracy of AI-generated analysis, 60% of respondents said cleaner data, 48% said visibility into how AI analysis is generated, 44% cited human feedback and moderation, and 40% cited tighter governance and security. Only 33% said they thought data accuracy would be improved by having better AI models.
The survey revealed that AI is also changing the roles and responsibilities of the data team. AI has fundamentally changed the data team’s core responsibilities for 86% of respondents, and 50% say that change has been dramatic.
About 20% of respondents said they’re looking outside the organization to hire additional data resources over the next two years, and 57% said they’re looking to upskill or reskill their current workforce to add AI-specific skills. Staff reductions were anticipated by 21% of respondents.
For C-suite technology leaders, the findings show that putting AI into production doesn’t mean teams will race to use it, and sometimes old habits are hard to change. Many employees will continue to rely on dashboards and manual reviews.
AI-based data analytics tools may provide answers, but employees aren’t yet willing to act on them.
Leave a Reply