AI in business intelligence: Uses, benefits and challenges

AI's role in business intelligence

Business intelligence applications, such as BI dashboards and interactive reports, have mostly been concerned with aggregating and analyzing historical data plus current information to get a picture of business operations. This is known as descriptive analytics: We describe the state of the business now and in the past.

However, it's only half the story that many business users want from their data. They also want data analysis systems that can tell them what might happen in the future -- predictive analytics -- and what to do about it, an approach sometimes called prescriptive analytics.

In the past, there was a big barrier to successfully deploying such systems: They're complex to build and, too often, to understand and act on. AI, with its analytical power and simplified user experiences based on natural language processing (NLP), can help shift the focus from descriptive insights to predictive and ultimately prescriptive ones. This enables businesses to foresee trends and take proactive actions.

AI-powered BI systems also enable real-time data analysis that provides decision-makers with more up-to-date insights. In addition, integrating AI into BI can drive business process automation, especially for simple or routine tasks. Doing so reduces manual effort and improves overall efficiency in companies.

Benefits of using AI in BI initiatives

The benefits of using AI in BI applications likely will be limited only by the imagination of BI developers and analysts. But some trends are already emerging. In particular, we can call out the following four benefits:

• Increased automation. The ability of AI to automate processes with minimal human intervention is an important economic benefit. Much of the work in BI involves data preparation, which can be quite repetitive or have well-defined outputs that are needed. AI can effectively automate both data preparation and data analysis, enabling business users in self-service BI environments to focus on strategic tasks that require their experience and business knowledge. This improves overall business productivity and should improve strategic performance too.
• Enhanced decision-making. Machine learning is a form of AI that can identify complex patterns even in vast amounts of data. Companies can use machine learning algorithms to run countless analytics experiments, exploring various scenarios, over the most complex data sets. That's most commonly done as part of more advanced data science applications, but the algorithms are now also being applied in BI initiatives. The result is a more thorough, more repeatable and often more insightful approach that helps lead to better business decisions.
• Improved business agility through real-time insights. AI's ability to analyze data rapidly and at scale enables businesses to respond quickly to market changes as they happen. Until now, most of the effort in real-time data analysis has been to ensure that source data streams are reliably and efficiently routed through an infrastructure of storage devices, data pipelines and analytics tools. These processes are often somewhat fragile. As a result, analyses must be kept relatively simple to not overcomplicate the data flow. AI enables more complex data handling with higher performance, making real-time business intelligence much more effective.
• Democratization of data analysis. As mentioned previously, the natural language query interfaces supported by newer AI tools enable a much simpler experience for business users working with BI data. They no longer have to learn query or scripting languages or even how to use data visualization tools. With AI -- in particular, generative AI (GenAI) -- users can converse with the BI system in natural language, using their business vocabulary. The analytics results are also easier to understand, and AI-driven augmented analytics features can generate visualizations and write code to run further analyses. This increased access to analytics capabilities for nontechnical users can help develop a genuinely data-literate culture in an organization.

Examples of AI applications in business intelligence systems

The benefits outlined above are all internal ones: They improve analytics and business processes. However, the greatest value of incorporating AI into BI initiatives likely will be in how companies can transform the customer experience (CX) and build businesses that are not just better run, but also more successful commercially. Here are some potential applications of AI in BI that will improve product and service strategies, CX, and future outcomes:

• Predictive analytics for market and consumer insights. AI can help businesses anticipate market shifts and customer behavior to guide strategic decision-making.
• Anomaly detection for risk management. Because AI systems can identify patterns in large and complex data sets, they provide early warnings for potential business risks, security threats and fraudulent activities.
• Sentiment analysis for customer service. The NLP features in AI tools have evolved significantly in recent years. For example, GenAI can not only generate high-quality text, but also analyze language features to understand emotions, concerns and needs. This sentiment analysis enables customer service interactions that are fine-tuned for individual cases, rather than reacting generically to questions and issues. Hopefully, that improves the customer experience overall.
• Supply chain optimization. Due to the effects of the COVID-19 pandemic, natural disasters and even container ships getting stuck in the Suez Canal, people are more aware than ever of the complexity and vulnerability of modern supply chains. This is a case where AI's ability to synthesize insights across the business is proving to be a compelling BI use. Companies can analyze supplier operations, logistics and patterns of customer demand in real time for more agile management of supply chain issues.

Challenges of implementing AI in business intelligence

While AI-driven business intelligence can help simplify many business processes, such advanced technology, which is still evolving, brings its own challenges. The following are some of the challenges that BI teams and business stakeholders should be aware of:

• Data management and governance. Although AI can automate data processes, it needs access to suitable data in the first place. The data a company currently uses for BI applications might be too highly aggregated for AI to do its best work on trend and pattern detection. If so, new data sets need to be prepared specifically for AI tools. Data also needs to be managed and governed effectively to avoid potential misuse. In the case of customer data or other confidential information, for example, concerns about data security and data sharing must be addressed.
• The black box problem. The complexity of AI models compared with BI processes can make it difficult to understand how they arrive at analytics conclusions. This often leads to concerns about the accuracy, consistency, fairness and transparency of AI-driven decisions.
• Ethical concerns and data privacy. The two previous challenges together lead to a third: AI implementation raises ethical questions about data privacy, bias and the responsible use of data. Customers need to be comfortable with the use of AI systems by organizations, and regulators increasingly demand strict compliance with rules on AI ethics and data usage. These challenges exist with conventional BI systems, but the increased autonomy of AI tools makes better analytics practices an urgent concern for companies.
• Skills gaps and the need for AI expertise. While AI can simplify work for business users running self-service BI applications, specialized skills are needed on IT and BI teams to design, deploy and maintain AI tools in BI systems. Companies might be able to upskill existing BI developers, analysts and administrators for this work, but highly skilled data scientists could be needed too. And in some cases, new employees might need to be hired to obtain the required skills.

Best practices for deploying AI tools in BI systems

The best practices for AI deployments are worthy of a detailed article in and of themselves. However, the following are some simple but important suggestions for integrating AI with existing BI practices:

• Align the AI in BI strategy with business goals. Technology of any sort isn't an end in itself. An effective AI implementation as part of the BI process begins with a clear understanding of the organization's overarching goals. Every tactical step should advance those business goals.
• Invest in data quality and data governance. High-quality data is crucial for successful AI integration, so it must be an upfront focus. However, ensuring both data quality and data protection in the long term requires a strong data governance program. In connection with that, companies should also consider creating an ethical framework for AI decision-making.
• Start small with pilot projects and scale gradually. Implementing AI in small, manageable projects enables experimentation and refinement before broader adoption. It also provides an opportunity for the organization to develop AI skills and find potential issues before making more substantial commitments.
• Upskill internal teams on AI. Creating a diverse team of BI professionals and business users with cross-functional expertise and domain knowledge is essential to ensure that AI use meets business needs. However, you don't need to hire all these people -- companies already have domain experts who can be trained on AI technologies and processes. The usability of modern AI is one of its big advantages. Training current employees also reduces internal resistance to the use of new AI technologies.
• Continuously monitor and improve AI deployments. AI is evolving rapidly as both a technology and a body of practice. For improved agility and accuracy, AI models in BI systems must be regularly updated. BI managers and their teams also need to stay on top of the latest AI developments to ensure that their processes, policies and skills keep up with the industry.

Future trends to watch for

Some users might think that AI will completely replace present-day BI. But it's more likely that AI tools will continue to augment BI software with new capabilities, while human insight remains a key component in the BI applications that drive strategic and tactical decision-making. The following are some of the further technical developments that we can expect over the next few years:

• Conversational analytics as the new standard. Natural language querying for analytics will become mainstream, simplifying interactions with data to resemble those with chatbots today. Query languages and tools for building data visualizations will be used only for the most advanced BI needs in specific use cases.
• Domain-specific AI models for industries. The development of domain-specific AI models will enable balanced insights that show a deep understanding of the business dynamics in different industries. As an example, AI-driven BI systems for the retail industry will understand the entire retail business process for insightful reporting and both predictive and prescriptive analytics.
• Automated and autonomous analytics. In addition to enabling business users to interact with BI systems using natural language, it's likely that AI tools will learn to look for patterns, anomalies and other insights in data sets and business processes without prompting. They might then deliver those insights to business decision-makers by automatically creating new data visualizations and dashboards.

AI is already revolutionizing BI applications by transforming data analysis from a retrospective process to a proactive, real-time one. The potential described here is real. The technical challenges are also real, but not insurmountable. The greatest barriers to increased AI adoption might be a lack of customer trust, employee resistance and the fallout from initiatives that are clumsily implemented.

But the guidance outlined here should give companies some insight into where AI in BI is going and how to sketch out a roadmap for successfully integrating AI technology with existing BI and analytics work.

Donald Farmer is a data strategist with 30-plus years of experience, including as a product team leader at Microsoft and Qlik. He advises global clients on data, analytics, AI and innovation strategy, with expertise spanning from tech giants to startups.