Business intelligence (BI) has historically been a platform for data and business analysts who build dashboards for executives to make decisions. But the generative AI (GenAI) explosion over the past 18 months has changed the game. 

People everywhere have tasted the power of ChatGPT and now want a natural language interface to every application. But can GenAI deliver powerful analytic data to anyone in the organization within any app?

Organizations face a significant challenge in distributing analytics across their entire environment due to the complexity of traditional dashboards and the limited expertise or time of many employees to handle data wrangling on their own. In a traditional BI environment, software licenses that enable users to author dashboards and visually explore data are typically limited to between 3% and 5% of the workforce. As a result, limited access for licenses to author BI reports and the need for specialized data analytics training become a chokepoint that prevents the democratization of organization-wide access to intelligence and data-driven insights. 

The cost and data analytics training required complicate broader adoption. What’s more, dashboards are largely static, with limited or no interactivity, and they serve only 20% or 30% of the workforce. Frontline roles are largely locked out of access to BI data.

Generative AI appears to offer a solution to this lack of data access. Many work professionals already cut and paste from ChatGPT or other large language models (LLMs) into business applications to save time in drafting emails, brainstorming a marketing campaign, or creating a project plan. Seemingly, every software vendor is announcing generative AI functionality. 

Unfortunately, generative AI alone is famously poor at crunching numbers, and it tends to hallucinate incorrect facts and figures, so it’s not a reliable foundation for data-driven decision-making. However, combining AI with BI can address the weaknesses that each technology possesses individually. GenAI expands BI to provide text explanations for data visualizations, answer natural language questions, and suggest wording for SQL queries, which can expand access. BI can provide generative AI with the data governance guardrails required to provide reliable intelligence. And, with the right design, this AI+BI combination can be integrated into the day-to-day apps of every employee, from the factory floor to the executive suite. 

Trusted BI can serve this function more effectively if it leverages a semantic layer. By itself, GenAI that searches internal data repositories will invariably find some good data along with data that is incorrect, outdated, or mislabeled. This is one core reason why GenAI models hallucinate: They struggle to differentiate between good and bad information. 

Did the sales team change their territory assignments recently? Chances are, there are some data sets that still reflect the previous sales territories. Did they realign territories the year before too? If the answer is yes, for a GenAI model, understanding what’s current versus what’s outdated or simply wrong is like the fairground game of three cups with a coin under one cup. To a data algorithm, each cup has slightly over a 33% likelihood of being right, so the algorithm picks one of the three at random to answer a data analytics question. For demanding enterprises, those odds are far too low for a winning business outcome. 

This is where the beauty of a semantic layer comes in for enterprise-grade AI+BI. 

A semantic layer technology vetted in production by demanding enterprises and public sector agencies enables authorized users with AI-powered assistance to create reports, dashboards, and visualizations without needing to know the underlying data structures. The semantic layer provides a consistent view of data across the organization, ensuring that everyone is using the same definitions and calculations, without the multi-year painful saga of previous-generation master data management initiatives. 

Ultimately, this type of governance will be a major factor in the success of your AI+BI initiatives. As longtime industry analyst and Gartner Distinguished VP Rita Sallam summarized memorably in her Gartner Predictions for 2024: Data & Analytics webinar on January 31, 2024, “Governance eats everything for lunch!”

Generative AI can liberate BI data from outdated static dashboards so that all employees have access, enabling them to make data-driven decisions. The magic begins with the combination of GenAI and a trusted BI platform built on the foundation of a semantic layer, where the BI and semantic layer have been tested for many years in production by demanding enterprises and government agencies.