Building trust into generative AI will revolutionise BI

Johannesburg, 26 Oct 2023
Craig Andrew, Head of Data Analytics, Keyrus.
Craig Andrew, Head of Data Analytics, Keyrus.

Generative AI is poised to revolutionise analytics and BI, but controls, governance and trust need to be built into the technology for organisations to embrace it.

This emerged during a webinar hosted by Keyrus and Tableau, in partnership with ITWeb.

Craig Andrew, Head of Data Analytics at Keyrus, said: “In a typical BI landscape, the types of data analytics being performed are descriptive – looking at past data to see what happened; then diagnostic – asking why something happened; and predictive – the ability to forecast. Prescriptive analytics has been the missing piece.”

“We have entered an exciting chapter of modern analytics, with BI and generative AI coming together to help us become more prescriptive and truly data driven.”

He highlighted Salesforce research which found that 57% of IT professionals believe generative AI is a game changer.

Taking flight with generative AI

Andrew said: “I believe generative AI has massive potential, but data professionals must pave the way for the ways in which it will change analytics.”

He noted that there were trust gaps around generative AI, with over 75% of consumers concerned about misinformation and 67% of organisations raising security concerns.

He said valid concerns included AI lacking the relevant data or context to give the right answer, resulting in ‘hallucinations’, as well as toxicity and bias.

“If we don’t trust generative AI, the prescriptive component of analytics won’t gain traction,” he said. “As with any new, innovative technology,we need to overcome the fear and help build trust in the technology.”

A poll of webinar participants found that 33% are definitely considering and 44% possibly considering deploying generative AI for their business needs in the next six months. On the question of what the most important factor was for enabling trust in their data strategy, 48% said data governance, 41% said data quality, 6% pointed to user training and 1% said transparent algorithms.

Andrew compared the emergence of generative AI to the early days of aviation, where trust and governance were needed to enable it to scale. Over the years, people grew to trust aviation with strict regulation and oversight, safety records and transparency, highly trained professionals, rigorous maintenance, crisis preparedness, and later international standards and positive experiences. AI needs to build trust in the same way, he said.

“We have seen disruptive technologies before as a society. With generative AI we need to learn from history to ensure people can trust it.”

To take flight with generative AI, organisations need to have a team of trusted experts and access to training and enablement, and trust their BI platform and its data governance capabilities, he said.

Building generative AI into Tableau

Francois Zimmerman, EMEA Field CTO at Tableau, outlined Salesforce’s new Tableau Pulse (Tableau GPT) generative AI for Analytics, which can anticipate the data each user needs based on their personal preferences and automatically generate actionable insights.

Zimmerman said: “We need to recognise that there are different types of data consumers out there – initially we at Tableau focused on the data analytics community. This type of user needs a data marketplace where they dig down until they find something relevant to them, and then look behind the data someone else generated to answer their specific questions. Historically, the experiences we brought into Tableau were optimised to support these data curious people.”

“However, there are many other people who spend time in their apps rather than their BI tool. Tableau Pulse makes a newsfeed-like experience for these users. It makes data easy for everyone. It's personalised, contextual and smart,” Zimmerman said.

“We’ve taken all the KPIs that matter most to these individual users, highlighting KPIs with the most interesting deviations from normal. These new experiences build on top of normal Tableau dashboards and data sources. Using AI-assisted follow up questions, users can dig deeper into the high level insights to discover what is driving the data points. In the next six months, we will launch unstructured search experiences, and this is where generative AI comes in.

“We also think users may need to share these insights in their collaboration channels, and other people may want to explore those metrics within their workflows, so Tableau Pulse will enable this. Behind the scenes ,we are building these experiences on top of generative AI models, using parent company Salesforce’s Einstein Trust Layer,” he said. “We use our deterministic insights platform and metrics layer to surface valid signals and our data cloud to stitch together different data versions into one canonical data model for a single reference with unified identities – so you get the entire picture of the customer.”

About Keyrus

Keyrus is your trusted partner in building sustainable, high-performing data architectures, utilising AI functionality and the latest data governance strategies and tools. We are ready to help you design an actionable data strategy that will deliver your business objectives and drive commercial success. Contact us at