The adoption of AI across the different verticals within an organization has produced a kind of ironic reversal in time. Despite this massive technological leap, the ability to create analyses, applications and tools is being decentralized within the organization, and that mirrors the proliferation of tools like Excel in the late ’90s and early 2000s. You get these precocious workers who are able to run interesting and potentially insightful analyses. But then you get data silos, where it’s not totally clear where they got their numbers from. Their methodology could differ from another team that performed a similar analysis and got a different result, and it becomes very difficult for business leaders to adjudicate the veracity of any one claim.
This creates two business problems. The first is the need, and the desire, to centralize analysis, which ironically puts greater importance on a highly governed, deeply manicured central data warehouse. The other problem is centralizing the analyses themselves, so that they can talk to each other and interrelate.
The organizations that figure out this puzzle first will reap the benefits. The ones that don’t will remain frustrated by a lack of progress in AI.