Advancing Data Literacy and Democratization With AI and NLP: Q&A With Qlik’s Sean Stauth

The boom in generative AI interest serves as a visible tipping point in the yearslong journey of the enterprise embracing the power of data interaction through natural language processing (NLP). In recent years, NLP has undergone significant changes that have made it increasingly easier for users at all skill levels to handle and explore data without being a data scientist.

The adoption of generative AI approaches is the latest example of NLP’s increasing potential to advance data literacy and democratization across the enterprise as well as drive performance for every employee.

Sean Stauth, global director of AI and machine learning at Qlik, discussed how NLP and generative AI can advance data literacy and democratization.

As director of AutoML solutions, Stauth helps Qlik’s global customer base develop and gain success with their AI and machine learning initiatives. A believer in the power of AI and predictive analytics to help companies with their strategic needs, Stauth has spent his career helping companies build AI- and data-driven products.

How can NLP and generative AI advance data literacy and democratization across the enterprise? Can you elaborate on this?

The largest barrier to widespread adoption of analytics within organizations is data literacy and the requisite skills. Not everyone is analytical or cares to spend time evaluating data for patterns and insights. Executives just want results, and managers often can’t afford the time needed to crunch numbers and thus make data driven decisions. NLP and generative AI change the game.

There are numerous examples of natural language interfaces being used by many people every single day. And in more recent years, NLP has undergone some significant changes thanks to advancements in machine learning and deep learning techniques.

[Take] a service chatbot, searching for a new item on Amazon, basically any Google search—we’ve all been using natural language to get answers and find items for years. Bringing that type of experience to enterprise analytics just makes sense, since it enables nontechnical staff to feel comfortable asking questions and trusting the answers they get back.

We know from experience that the more someone uses any service or technology, the more comfortable they become. That’s part of what’s driving this incredible interest in generative AI. The interface is so simple to use, and the results are easily understood, that there’s really no skill gap to overcome.

Because of this, as organizations start to bring generative AI into more workflows, staff will naturally start to use those services more regularly, exposing them to more data and using data more regularly, creating a positive cycle that will only build on itself.

Why is this so important?

Despite continued investments over many years, according to Gartner, 85% of data analytics projects fail. There are various reasons for this, but one key ingredient is the lack of data skills across the business. We often see an enterprise deploy analytics to different parts of the organization, without coupling that with skills training. Inevitably, they don’t see the adoption they were hoping for. The result is that despite the investment, staff are making decisions without key data, which leads to decisions that aren’t as strategic or impactful.

This also increases the risk of business units being left behind and an increasingly stark parallel of business opportunities being lost because of it. Generative AI changes the playing field. With simple, intuitive interfaces, the adoption can move beyond technical departments.

Raising the level of data use and comfort with data through generative AI interfaces will mean more people actually using data in everyday decisions.

What is Qlik doing to help companies achieve data literacy and democratization?

We’ve long been a champion of data literacy as a founding member of the world’s first data literacy project, with leading organizations such as Accenture, Cognizant, and Experian. We’ve also provided a wide range of data literacy training courses for free to both professionals and academic institutions to help anyone who wants to become more skilled to do so. This is reinforced in our products as well. We’ve had natural language interactions, search, and AI-powered insights integrated directly into our solutions for years to make it easier for any Qlik user to find answers, explore their data, and discover hidden insights. And a core focus of our R&D efforts is simplifying the adoption of technologies such as machine learning. Our AutoML capability is purpose-designed for business analysts and doesn’t require previous expertise in data science or machine learning.

How would you address some of the concerns regarding NLP and generative AI?

The biggest issues we see right now with generative AI are driven by data quality and governance. Trust is key. But any organization that tries to shut down use of generative AI due to risk is kidding themselves—people are going to use it no matter what, given how easy and powerful it is. Organizations need to be proactive in identifying the areas where generative AI can bring value. At the same time, they need to audit their data framework and set up the right data quality and governance processes. Governance ensures core enterprise data is not being used outside the four walls. Data quality keeps you from feeding incomplete or biased data to the algorithm, which is crucial in reducing the hallucinations everyone is hearing about. Simply put, there is no generative AI without data—it’s all about the data, but it has to be the right data. So, getting your “data house” in order is where it all begins.

How do NLP and generative AI assist data engineers and analysts? Are there other job roles that will benefit?

We are seeing many instances where NLP and generative AI are helping developers augment their efforts with code generation, taking out hours of manual time that they can then apply to other tasks. It can massively accelerate previously mundane tasks like data discovery and preparation. Similarly, analysts can more quickly explore data for what-if scenarios, especially when using NLP or generative AI as a layer on top of an AutoML solution for predictive analytics efforts.

What does the future hold in this area?

We’re only in the early days of seeing what new applications will come from generative AI and NLP. We see the market quickly heading to enterprises deploying private instances of these technologies with large and small language models within their own networks, instead of using open public platforms to gain control and avoid risk. We also see an increase in adoption of data quality and governance—since these are crucial in making sure the algorithms are being fueled by trusted, relevant, and unbiased data so users can trust the outcomes and make more confident and certain business decisions.


Subscribe to Big Data Quarterly E-Edition