Hearing the vendor hype, enterprise buyers might think they’re all in when it comes to generative AI. But as with any new technology, large companies tend to tread carefully. Throughout this year, CIOs have been paying attention as vendors have eagerly announced new generation products powered by his AI.

Some companies are actually looking at reducing spending, or at least smoothing out spending, and are not necessarily looking for new ways to spend. The big exception is when technology allows companies to operate more efficiently and accomplish more with less.

Generative AI certainly has the potential to do that, but it either increases the cost of these features in a SaaS product or the cost of using a large language model API if you build your own. It also comes with its own costs, such as how much it costs. Software internally.

Either way, it’s important for those implementing the technology to understand whether they’re getting a return on their investment. Many companies are proceeding cautiously, with 56% of respondents reporting that generative AI is impacting their investment priorities, according to a July Morgan Stanley survey of CIOs of large companies. However, only 4% of people actually launched any significant projects. In fact, most were still in the evaluation or proof-of-concept stage. This may be a rapidly changing area, but it’s also consistent with what we heard in our conversations with CIOs.

That said, similar to the consumerization of IT a decade ago, CIOs are under pressure to deliver the kind of experience people see when they play ChatGPT online, says Madrona Ventures Partner says Jon Turow.

“I think it’s undeniable that all of our corporate employees, who are internal customers of CIOs and CTOs, have tried ChatGPT and know how great it is. , and know where the great words are. So CIOs are under pressure to achieve that level,” Turow told TechCrunch.

Particularly where some of the pressure may be coming from the CEO, the desire to please internal customers and potentially transformative things like generative AI. There is also a tension between CIOs’ natural tendency to act cautiously. Jim Rowan, a principal at Deloitte, said that making this happen requires building some structure and organization over time, and how to build generative AI across the enterprise in an organized way. He said he is working with customers.

“A lot of the way we work with companies is to think about what infrastructure they need to be successful. Infrastructure doesn’t necessarily mean technology, but people. Who is that, what is the process and governance…and giving them the ability to set it up,” Rowan said. A big part of that is talking about use cases and how the technology can be used to address specific problems.

This is consistent with how the CIOs we spoke to are implementing this in their organizations. Monica Caldas, her CIO at insurance company Liberty Mutual, started with a proof of concept for a few thousand people and is looking for ways to scale it at her 45,000-employee company.

“We know that generative AI will continue to play a critical role in virtually every part of our company. We are investing in use cases to further develop and refine them,” she said. she said.

Mike Haney, CIO of Battelle, a science and technology-focused company, is also exploring generative AI use cases this year. “So we’ve been working on advancing AI for the past six to nine months, and we’re currently building out specific use cases for different teams and functions within the company.” Although it is still early and they are still exploring ways in which it can help, they caution that so far the results have been good in terms of providing more efficient methods.

Kathy Kay, executive vice president and CIO of financial services firm Principal Financial Group, says her company started from scratch with a research group. “So we opened it up to any employee with an interest or passion, and the number grew to about 100 people. It’s a combination of engineers and business people, and now she’s probably working on 25 use cases. of which she plans to put three into production. [soon],” she said.

Sharon Mandel, Juniper Networks CIO, said her company is participating in an early pilot with Microsoft for Copilot for Office 365, and anecdotally, some people like Copilot and others are less impressed. says they’ve heard mixed feedback. Measuring productivity gains remains a challenge, he said, even though Microsoft has started offering dashboards that at least show levels of adoption and usage.

“The difficult thing about this is that we don’t have data on people’s productivity levels. So no matter what, we want to make sure that we have a good understanding of Microsoft’s dashboards that show how our users are using them. Until then, we will be using somewhat anecdotal information,” she said.

When companies hear about the potential power of generative AI, it’s no surprise that they want to learn more about it and leverage it to make their organizations run more efficiently, but at the same time, executives are becoming somewhat cautious. Of course. We recognize that these are still in their early stages and we need to learn through experimentation whether this is truly a revolutionary technology.