1.6 AI in Business Intelligence

Artificial intelligence is fundamentally reshaping how organizations approach Business Intelligence (Dwivedi et al. 2021). Where traditional BI relied on analysts manually querying databases, building reports, and interpreting results, AI-augmented BI systems can automate many of these steps — surfacing insights that might otherwise go unnoticed and making sophisticated analysis accessible to a broader range of users (Davenport and Ronanki 2018).

1.6.1 What is Artificial Intelligence?

Artificial intelligence (AI) refers to computer systems that can perform tasks typically requiring human intelligence — such as recognizing patterns, making predictions, understanding language, and generating content. Unlike traditional software, which follows fixed rules written by a programmer, AI systems learn from data. A machine learning model, for example, does not need explicit instructions for how to detect fraudulent transactions; instead, it examines thousands of historical examples and learns to distinguish legitimate transactions from suspicious ones. This ability to learn from data rather than follow predetermined rules is what makes AI particularly powerful for BI, where the patterns of interest are often complex, subtle, and changing over time.

AI is not a single technology but a broad family of techniques. For BI practitioners, the most relevant branches include machine learning (algorithms that learn from data), natural language processing (systems that understand and generate human language), and generative AI (models that create new content such as code, text, and visualizations). Each of these will be explored in more detail below and throughout this textbook.

1.6.2 AI-Driven Analytics and Automated Insights

Modern BI platforms increasingly incorporate AI to move beyond descriptive reporting toward predictive and prescriptive analytics (Loureiro et al. 2021). Machine learning algorithms can analyze historical data to forecast sales trends, predict customer churn, or identify operational bottlenecks before they become critical problems. Some platforms now offer automated insight generation, where the system proactively identifies statistically significant patterns in the data and presents them to users in plain language, without requiring the user to formulate specific queries.

1.6.3 Natural Language Querying

One of the most visible applications of AI in BI is natural language processing (NLP), which allows users to ask questions of their data in everyday language. Rather than writing SQL queries or navigating complex dashboard interfaces, a marketing manager can type “What were our top-selling products in the Midwest last quarter?” and receive a formatted answer with supporting visualizations. This capability dramatically lowers the technical barrier to data-driven decision-making, enabling more employees across an organization to engage directly with BI systems.

Tools like Power BI’s Copilot, Tableau’s Ask Data, and open-source text-to-SQL models translate natural language questions into database queries automatically. An analyst who previously needed to write a complex SQL join across three tables can now describe what they want in plain English and receive both the query and the result. For students learning BI, this means you can start asking meaningful questions of data from day one — and then examine the generated code to understand what happened under the hood.

1.6.4 AI, Coding, and the Modern BI Practitioner

Coding is a foundational skill in BI because it gives practitioners direct control over data manipulation, statistical analysis, and visualization. GUI-based tools like Power BI and Tableau are powerful, but they constrain the analyst to the operations the interface provides. Code — particularly in languages like R and Python — allows analysts to handle messy data, build custom models, reproduce analyses exactly, and automate repetitive tasks. Understanding code also builds transferable knowledge: an analyst who understands how a regression model works in R can evaluate the output of any tool that runs one, whether it is Tableau, SPSS, or an AI assistant.

AI tools are now transforming the coding experience itself. Large language models can generate R code from a plain-language description (“build a logistic regression predicting high absenteeism using age, BMI, and commute distance”), explain what existing code does line by line, debug errors, and suggest improvements. GitHub Copilot auto-completes code as analysts write, while AI-powered notebooks can execute code, display results, and provide narrative interpretation in a single workflow (Cui et al. 2024; Ziegler et al. 2024).

This does not eliminate the need to understand code — it raises the stakes. When an AI generates a 20-line R script, the analyst must be able to read it, verify that the logic is correct, and recognize when the AI has made an error. A practitioner who cannot evaluate AI-generated code is more dangerous than one who writes code slowly, because they may confidently present flawed analysis. Throughout this textbook, we will use AI tools alongside R to demonstrate how the two work together: AI accelerates the work, while conceptual understanding ensures the work is sound.

1.6.5 Key AI Concepts for BI Practitioners

Several AI concepts are particularly relevant to BI practitioners:

  • Machine Learning (ML): Algorithms that learn patterns from data and improve with experience, used for prediction, classification, and clustering tasks in BI.
  • Natural Language Processing (NLP): The ability of computers to understand and generate human language, enabling conversational interfaces to data systems.
  • Generative AI: AI systems that can create new content — including code, text, and visualizations — based on patterns learned from training data. Large language models (LLMs) like ChatGPT and Claude are prominent examples.
  • AutoML: Automated machine learning platforms that handle model selection, hyperparameter tuning, and feature engineering, making predictive modeling accessible to analysts without deep ML expertise.

These concepts will be revisited throughout the book as we demonstrate their practical application in BI workflows.

Example: AI-Assisted Data Exploration

Imagine you receive a new dataset — 10,000 rows of employee records with 20 columns — and need to understand what it contains. Traditionally, you would write R code to examine the structure, compute summary statistics, and generate exploratory plots. With an AI assistant, you can take a different approach:

  1. Upload or describe the dataset to an LLM (e.g., ChatGPT, Claude, or GitHub Copilot in RStudio)
  2. Ask: “What are the key variables in this dataset? Are there any missing values or outliers?”
  3. The AI generates R code to answer your question, explains what the code does, and summarizes the results in plain language
  4. You review the code, run it, and verify the output against your own understanding

This workflow does not replace the need to understand R or statistics — you still need to evaluate whether the AI’s code is correct and its interpretation is sound. But it dramatically accelerates the exploratory phase and helps you ask better questions faster. We will use this approach throughout the case study chapters in this book.