Organisations need clear data policies and standards, solid data governance and robust security protocols in place before attempting to scale AI within the business.
This is according to technical experts from InfoVerge, who were speaking during a webinar in InfoVerge’s Digital Frontier series of webinars. They outlined the importance of quality data and data governance in building a foundation for AI.
Ntiyiso Mayile, Chief Data Officer at InfoVerge Solutions, noted that the AI landscape was vast and complex. “As much as AI offers great things, one thing remains certain – underneath it all, it doesn't just happen. Fundamental data inputs are required.”
Skhumbuzi Mjoji, CTO at InfoVerge Solutions, said: “As AI rapidly reshapes industry, one thing is clear – without high quality, well governed data, even the most powerful AI cannot deliver trusted, scalable outcomes.”
Noting that AI requires readily available structured, semi-structured and unstructured data for training purposes, Mayile agreed: “Your AI models are only as good as the data on which they are trained, so you need a strong data foundation.”
Mayile said many organisations faced significant roadblocks in the way of scaling AI to production. “There has to be very good data in place, or everything will go pear-shaped,” he said. Among the roadblocks were a lack of a proper data foundation and strategies to scale AI pilots to production, a lack of data governance, fragmented technologies and data silos, and poor data management. “Without appropriate data management and governance to protect and control the data, there are risks associated with the use of AI, including bias, security vulnerabilities and unreliable outcomes. So AI could become a liability,” Mayile said.
“In many cases, challenges in harnessing AI start with a lack of data culture and data not being ready for organisations to seriously take advantage of AI models,” he said. “You need a lot of quality data and a lot of compute. Another challenge is many organisations don't have clear objectives for what they want to use AI for. To become an AI-first organisation, you need to start treating your data as an asset, not just as a by-product of your business processes. You also need to invest in the necessary infrastructure to support the AI workflow, rather than trying to retrofit infrastructure that is not scalable.
“Organisations also need to achieve a balance between AI innovation and governance, with a flexible governance framework that adjusts to potential impact and promotes cross-functional collaboration and transparency among all teams,” he said.
He also highlighted the importance of developing a data culture within the organisation: “Data culture is not IT’s responsibility – it’s everyone’s responsibility. We need to make everyone across the organisation understand that every piece of data they touch is important.
“By understanding that your data is no longer just for reporting or traditional business intelligence, you must intentionally design the data strategy to fuel the outcomes from our processes. This starts by clearly defining your AI use cases. You cannot just adopt without thinking about what you want to achieve,” Mayile said.
Mjoji added: “We see people doing amazing things with AI, but there has to be a lot of consideration before going full scale into AI.”
The big question facing organisations now, said Mayile, was how to establish a robust, scalable and trustworthy AI foundation while effectively leveraging AI data pipelines, mitigating risks and ensuring responsible use of AI.
He said InfoVerge recommended five key pillars for a robust data foundation to enable organisations to take advantage of AI as it grows. Organisations need to build an AI-first data strategy, developing a robust data strategy and prioritising AI integration and access to the organisation. They need to create a unified, reliable data repository for consistent and accurate access to data; and to fast-track the process of leveraging AI initiatives to drive faster insights. They should also manage, govern and secure data across the AI life cycle, and should streamline data operations to lower costs and facilitate scalable AI deployment.
InfoVerge’s webinar series will also include an event entitled: ‘From Data to Intelligence – Activating Smarter Operations’ on 23 July; and ‘Scaling AI Success – Governance, Ethics, and Future-Proofing’ on 17 September. The webinars will highlight the tools and practices required to scale and future-proof AI initiatives.
To learn more and register for these webinars, go to:
Share