Arun Ulag, Microsoft corporate vice president for Azure Data, reckons that, in a world of constrained or finite IT budgets, something will have to give if new projects are to thrive.

Speaking at the Citi 2024 conference recently, Ulag was outlining thoughts on roadblocks and constraints customers face in the latest data modernization round promoted by the tech giants.

As well as noting the complexity of building generative AI applications, he also highlighted that budgets were insufficient to support overall investment in all the fresh projects arising from the technology.

"If this thing goes up a lot, something else has to adjust," he said. "Customers are looking for savings and making sure they can fund these initiatives."

In other words, if you want shiny AI baubles, something else will need to be sacrificed.

Ulag's comments mirror the findings of a study conducted by the IBM Institute for Business Value, which found that while IT bosses were concentrating on getting ready for generative AI, executives suspected basic IT services were being neglected.

From Microsoft's point of view, if enterprises would only deploy more Copilots onto desktops and roll more AI into their technology stacks, the cost savings and productivity improvements would more than pay for all of those new projects.

But perhaps not. Ulag also emphasized the risks of rolling out AI services. "LLMs are new," he said. "They behave in different ways, often unexpected."

Ulag gave the example of using an LLM for customer service and having to ensure the content produced was not the result of an AI hallucination or liable to cause offense.

Microsoft's solution for this is investing in responsible AI, but a glimpse at the company's recent Copilot for Microsoft 365 Transparency Note indicates that it still has some way to go.

In its note, Microsoft says: "We encourage users to review all content generated by Copilot for Microsoft 365 before putting it to use."

As Ulag says, the technology is new and might do something unpredictable. Although this can be somewhat mitigated by increasing the grounding to include more corporate data, it remains a risk.

Back in March, Microsoft admitted it was still trying to walk customers by the hand through proof-of-concepts to convince them of the producitivity benefits of $30 monthly costs per seat for Copilot for M365.

Preparing an organization for the deployment of AI technologies is an expensive and time-consuming business. While Microsoft and its peers might point to the potential for productivity gains and cost savings - who needs staff when you have robots? - getting there will require investment. ®

https://www.theregister.com//2024/09/09/microsoft_arun_ulag_ai/