Generative AI is the latest technology to shake up data analysis, a field with a long tradition of combining advancements in technology with new ways of doing business. Data analysts should adhere to best practices for generative AI use in analytics operations if they want to reap benefits.
Data analysis evolved from decision support. It developed into data warehousing, BI, visualization and predictive analytics. At each step, decision-makers demanded better insights, so IT learned to deploy and manage challenging technologies. As a result, companies found new efficiencies or entirely new commercial models.
Generative AI is the next evolution of data analytics — and data operations as a whole — but it’s not clear what effects it will have. Analysts must adapt to a constantly evolving field of opportunity to learn how to take advantage of the new capabilities AI offers.
The key to the power of generative AI is in the name. Powered by machine learning models, generative AI can create text, images, code, new data and more. Many people are familiar with AI-created images and text, but generative AI is equally adept at analyzing information, a capability stemming from its training methods.
During training, generative AI learns to identify patterns, predict outcomes and extract key features from vast data sets. The same techniques applied analytically enable the AI to interpret and analyze new data effectively. The dual capability of generation and analysis makes generative AI a uniquely versatile technology.
Generative AI uses for data analytics
Organizations can demonstrate the usefulness and efficiency of generative AI in data analytics in several ways.
Generating synthetic data for analysis
The lack of good quality, openly available data is a common barrier to building and testing new analytics tools and effective machine learning models. Available data is often limited in scope and does not reflect the complexity of real-world scenarios.
For machine learning, it is difficult to find data that contains rich and realistic patterns. Even if one data set fulfills the desired data needs, it can be a challenge to find others for testing and validation. In all cases, the use of real data poses ethical challenges related to the privacy and security of confidential data.
As a result, the same data sets get used repeatedly, such as the numerous demos from software vendors all replaying the same analytics over data from the Olympic games, taxi data from New York or movie rentals.
Today, generative models can create vast data sets of synthetic — yet realistic — data to fuel analytics and modeling initiatives. Synthetic data serves two crucial roles. First, it addresses privacy concerns in data analysis, particularly in sensitive sectors such as healthcare, by creating realistic, but non-real data, thus protecting individual privacy. Second, it fills gaps in scenarios where actual data is scarce or nonexistent, such as unique market trends or emergency situations. The simulation of rare scenarios allows for more comprehensive modeling and analysis, which significantly enhances the usefulness and pertinence of data-driven insights. The result is more interesting and meaningful analytics for data analysts.
Uses in enterprise BI
By generating charts, summaries and dashboards, generative AI has the potential to automate routine BI reporting. The same technology can also identify patterns that human analysts or business users miss and explain insights in natural language. Automation frees up data analysts to focus less on rote tasks and more on higher-value analysis.
However, generative AI capabilities go beyond reporting. Traditional BI focuses on descriptive analytics, summarizing and interpreting trends in historical data, providing insight into what has happened. In recent years, predictive analytics has become mainstream, using statistical algorithms and machine learning to suggest future trends or what may happen.
Generative AI makes prescriptive analytics possible and practical. Prescriptive analytics provides advice on predicted outcomes, recommending actions, tactics and strategies based on predictions. Human analysts and strategists, working with the prescriptions, can be more perceptive, confident and innovative.
Generative AI’s creation capabilities improve data analytics usefulness.
Benefits of using generative AI in data analytics
It seems likely that generative AI will redefine the landscape of data analytics, in time. However, just as data warehouses are still fundamental to enterprise architecture more than 30 years after their initial development, expect today’s methods for analytics and reporting to be in use for years to come. Generative AI has potential benefits, not just as a technology, but as an enhancement to existing analytics techniques and tools.
Increased automation
Generative AI’s ability to find patterns and trends even in complex, messy data reduces the need for manual data processing, leading to cost savings in labor and time. Instead of working on data labeling, cleansing and normalization, human experts can shift their focus to strategic, high-value work instead. Automating mundane, repetitive tasks also ensures consistency; manual cataloging is fallible. Automated reporting and analysis enable organizations to make decisions faster based on more up-to-date data, which supports more agility across the enterprise.
Identifying patterns, correlations or relationships
Generative AI excels at identifying complex patterns, correlations and relationships in data that human analysts might not see. Generative AI can simulate different scenarios to identify risks before they happen, allowing businesses to proactively develop mitigation strategies. It can also identify prospects for growth, such as new markets, products or services.
Human analysts and strategists, working with [generative AI-supported] prescriptions, can be more perceptive, confident and innovative.
For example, a financial institution could use generative AI to replicate patterns from real financial transactions along with new similar patterns to train fraud detection models. The capabilities of generative AI improve the ability for the organization to discern fraudulent trends and enable new financial products that are safer and more aligned with realistic consumer needs and behaviors.
Efficient data catalogs
A data catalog is an organized inventory of data assets, which can discover and provide relevant data to users with the right permissions. A good catalog offers fast and self-service access to appropriate data with meaningful context. Generative AI can automate the cataloging process, and it can intelligently categorize and tag data sets, which makes the catalog more usable. Automation also ensures data quality and consistency, which is crucial for better data governance and management.
Generative AI and data analytics best practices
As with any new technology, best practices for generative AI are developing as fast as the tech itself. Nevertheless, some of the basic guidelines should be useful in any implementation to maximize benefits.
Use high-quality data
Generative AI excels at identifying patterns in complex data and can generate new data sets, but its effectiveness in prediction, pattern detection and automated decision-making relies on the quality of the input data. High-quality business data enables generative AI to produce reliable and accurate results. Data cleansing, quality control and data governance are core investments for any organization using generative AI.
Integrate tools with generative AI
BI tools are catching up with generative AI. Tools that integrate generative AI with existing data infrastructure simplify adoption and streamline workflows. Organizations can choose between data analytics platforms with built-in generative AI capabilities or tools that integrate generative AI to enhance their existing data analytics operations.
Determine KPIs, goals and use cases
Setting clear goals in the form of KPIs or Objectives and Key Results before starting with generative AI is a useful step to manage the technology effectively. Consider who might use the tool, any industry requirements, cross-department uses, presentation formats, the speed or rhythm of the business, the accuracy required and the training needs of human users.
Tailor to specific goals and needs
Designing generative AI implementations and integrations for specific scenarios ensures the most effective use. Whether it’s enterprise BI, marketing, sales, customer experience analytics or geospatial analytics, customizing generative AI resources maximizes their potential, rather than relying on generic models which may have limited understanding of the unique contexts and nuances of different industries.
Generative AI has already transformed so much in data analysis, presentation and operations. As the technology matures, it should continue to fundamentally alter how companies build value from their data assets.
Start experimenting with integrative applications of generative models, particularly in some of the use cases described. The potential for enhanced decision-making through automation, deeper insights and increased efficiency is genuinely exciting. Analytics teams willing to take on the challenge have an opportunity to dramatically change their own role and even the fundamentals of the business. It’s a unique and inspiring prospect.
Donald Farmer is the principal of TreeHive Strategy, who advises software vendors, enterprises and investors on data and advanced analytics strategy. He has worked on some of the leading data technologies in the market and in award-winning startups. He previously led design and innovation teams at Microsoft and Qlik.
