By using website you agree to our use of cookies as described in our cookie policy. Learn More

TDWI Upside - Where Data Means Business

How to Transform Your Business Through AI

James Kobielus, TDWI’s senior director of research for data management, discusses how to transform your business through AI.

In this “Speaking of Data” podcast, TDWI’s James Kobielus explores how AI can transform your business. Kobielus is senior research director for data management at TDWI. [Editor’s note: Speaker quotations have been edited for length and clarity.]

For Further Reading:

How the Right Data Management Foundation Fosters Digital Transformation Success

Five Best Practices to Digitally Transform Your Business

Four Steps to a Successful Digital Transformation

“AI transformation is just a subset of the larger digital transformation of business that’s been going on for quite a while now,” Kobielus said. “Digital business processes have become the core of the modern economy as consumers and users conduct more of their lives through digital channels.” Businesses are using these digital capabilities to improve customer experience and retention, he added.

Kobielus also noted that as businesses consolidate more processes down to fewer platforms and scale them to encompass a greater number of users, their own internal operations and data-driven decision-making can be improved.

He explained that there are a multitude of methodologies that fall under the general umbrella of “AI” -- for example, machine learning, deep learning, neural networks, and large language models. Their essential common feature is that they automate the need for human judgement and supervision over a wide range of intelligent processes, such as customer interactions or managing the backend supply chain.

“What most people think of nowadays, though, are the generative AI tools that the media is going wild over,” Kobielus noted. “These are the tools such as ChatGPT and Midjourney that can write text or create images from a text-based prompt.

“Another thing most of these technologies have in common is that they are cloud-based,” he explained. “AI is extremely resource-intensive in terms of the processing resources and storage it uses, in addition to the bandwidth required to communicate its insights.” This also enables organizations to use modern development practices such as DevOps and continuous development/continuous delivery processes to produce and deploy pipeline code.

Given how easily an organization can sink huge amounts of resources -- both financial and otherwise -- into AI initiatives, Kobielus offered several areas on which to focus.

“A key focus always needs to be governance,” he said. “That means governing both the data -- making sure it’s cleansed, corrected, and deduplicated -- as well as the AI models that the data is used for.” He also pointed out the importance of having a range of development tools geared toward a variety of personas in the organization, such as data science workbenches for the professional data scientists and self-service tools for the non-traditional developers and business analysts who are increasingly becoming the next generation of analytics developers.

“It’s also important, when using these new AI tools that there be a human in the loop,” Kobielus said, “to help reduce the chances that they will turn out offensive or biased output. Such output can very easily cause serious damage to your reputation or cause legal or regulatory woes. That also includes hallucinations -- false or misleading output.”

He pointed out that there are some recent approaches such as retrieval-augmented generation (RAG) being applied to the problem that are expected to help reduce issues such as hallucinations. However, organizations will only be able to tell what approaches will work for them by trying them.

A related risk of the new generation of generative AI tools is that users will be asking questions about topics they don’t understand well enough to evaluate the output they receive, Kobielus said. “Some of this can be addressed by careful prompt creation,” he said, “but this is where having a subject matter expert in the loop can be vital.”

Kobielus wrapped up by mentioning that TDWI is currently surveying data and analytics professionals to solicit their opinions on where their transformation efforts stand today and the degree to which they’re using AI to achieve them, and invited readers to participate.

TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, and Team memberships available.