As interest in AI development surges, the underlying data used to train models and applications becomes increasingly important.
As a result, 2025 will likely be a year when enterprises place greater emphasis than ever on the basic tenets of proper data management -- how it's governed, stored, prepared and analyzed - according to Tony Baer, principal at dbinsight.
"I see 2025 being the year of the renaissance of data," he said. "As AI projects get closer to production, enterprises will start to pay attention again to data."
Enterprise interest in developing AI-powered applications has grown exponentially in the two years since OpenAI's launch of ChatGPT marked a significant improvement in the capabilities of generative AI (GenAI) models. Large language model technology has improved since then, with AI developers Anthropic, Google, Meta, Mistral and others all striving to top one another.
But without an organization's proprietary data, those models are of little use.
It's only when GenAI is combined with proprietary data and trained to understand an organization's operations that the models become useful. Only then can they deliver benefits such as smarter decision-making and improved efficiency that make generative AI such an attractive proposition for businesses.
So as enterprises invest more in AI development, they will also need to take steps to ensure that their data is properly prepared.
Data lakehouses and data catalogs will be front and center. So will data and AI governance. And accessing and operationalizing unstructured data will be critical.
In those organizations that have their data to properly develop AI-powered applications, traditional business intelligence will be transformed, according to Yigal Edery, senior vice president of product and strategy at Sisense.
"In 2025, AI will completely obliterate the boundaries of traditional BI, enabling anyone to develop and use analytics without specialized knowledge," he said. "Emerging AI-driven platforms will make analytics as intuitive as natural conversation, eliminating the need for clunky dialog boxes and complex interfaces."