A new study of nearly 200 enterprises that have already adopted some form of generative artificial intelligence (GenAI) found they have judged GenAI to be a strategic priority. As such, many have C-suite leaders involved as part of a centralized strategy of testing, deployment, and governance. But that doesn’t mean these companies haven’t been pushing the boundaries on use cases.
The study by Computer Economics (a service of Avasant Research), Generative AI Strategy, Spending, and Adoption Metrics, found that 40% of the companies had put the chief information officer or IT steering committee in charge of the GenAI effort, and another 34% put the CEO or board of directors in charge. Cross-functional teams were in charge at 14% of the companies, while CFOs had lead authority at 6% of the organizations and chief risk officers at 5%.
“Having a few early adopters using GenAI individually might lead to a couple of short-term wins, but having a defined set of tools, strategy, governance, and budget can lead to sustainable boosts in productivity and new revenue,” according to the Avasant report.
Budget-wise, centralization also appears to be the common approach. More than four in 10 organizations (44%) said their GenAI budget was a component of their central IT budget. Another 40% said GenAI spending was housed within a different budget, sometimes within a single business unit or department.
For companies whose GenAI spending was within the central IT budget, the median spending was 4% of the budget, a significant number for such a young technology.
David Wagner, director of research at Avasant, said it’s too early to use that number as a benchmark because some companies were only spending a couple of thousand dollars on pilot programs or fractions of a percent of their IT budgets. But he also said it was a surprise to find some companies far enough along that they had seven-figure budgets.
Where will the money come from? While some companies will be aggressive and invest significant dollars upfront, Wagner thinks the “zero-cost transformation” method will be popular, as it was with cloud computing. In zero-cost transformation, the dollars generated by productivity enhancement, once realized, are re-invested back into other cost-savings or revenue growth areas. “So that you’re never actually paying for any of this stuff; it’s paying for itself,” Wagner told CFO. “I suspect a lot of that is happening in GenAI.
While it may be easy for companies to see possible efficiency gains from GenAI, that can be limiting. Gen AI can also boost revenue by being used in product design, prototyping, and finding the market value of an idea — and in general help employees be more creative. Nestlé uses a proprietary GenAI tool to validate new product ideas based on the tastes and preferences of existing products.
Avasant found some exciting use cases and pilot programs occurring inside enterprises. Two companies were using AI chatbots to handle first-round interviews with job candidates. A retailer with only a small business-to-business (B2B) sales component takes B2B orders solely through a GenAI portal. A professional services company replaced their entire procurement team. And a financial services company turned over all writing of employee performance reviews to a GenAI tool.
While these use cases are impressive, they’re not predominant. The biggest problem with GenAI right now, said Wagner, is understanding “the art of the possible” — understanding how to govern it, what to be afraid of (cybersecurity, factual errors, results bias), and figuring out the use cases in a given industry. (See chart above.)
These organizations already in the throes of experimenting with GenAI see the technology creating jobs not taking them away. One-third of the business leaders surveyed (34%) said it would lead to major job creation, and another 48% said it would create “some” new jobs.
According to Wagner, less-experienced workers have the largest productivity gains from using GenAI. “Senior folks aren’t gaining as big a benefit because they already have the knowledge contained inside the AI,” Wagner said.
While GenAI has its flaws, it’s here to say, he said. “We're going to have to learn to govern it. We're going to have to learn its limitations. [But] everyone can learn from it and use it.”