Like big data in 2013, we are in the “everyone's doing it, but no one knows why” stage of generative AI (genAI). A recent McKinsey survey found that 65% of companies “use genAI regularly.” It's exciting! In Elastic's recent earnings call, the company said that over 1,000 customers are paying to build genAI applications. Amazing! It's not just Oracle, each of the major cloud companies is telling us that genAI is driving their cloud spending. Amazing!
Maybe not.
Peel back the headlines and we see genAI as merely aspirational, not necessarily transformational, for most companies. For example, Elastic CEO Ash Kulkarni, while touting all the customers building genAI applications, also said, “We don't expect genAI to be a significant revenue contributor this year.” In other words, the 1,000 companies aren't doing much, so they're not paying much. This is not a disrespect to Elastic, but the current state of genAI today: the cloud is inflating AI revenue primarily through training models, not companies using those models to draw inferences from data in their applications.
In other words, if you haven't yet transformed your business with AI, you're not alone — there's still time.
It's too early for genAI
I wrote about this recently, but won’t repeat myself (i.e., companies that are truly successful, not big genAI projects, tend to do better search through search augmentation generation (RAG).) According to a McKinsey survey, companies still aren’t sure exactly where to use genAI. Only two use cases were cited by more than 15% of respondents: “content support for marketing strategy” and “personalized marketing.” There’s also IT help desk chatbots (7% of respondents) and design and development (10%), but for the most part, everything else is an estimate.
Companies are experimenting, to say the least.
Other data from the survey raises more questions than it answers. For example, the report notes that “respondents most commonly reported a significant increase in revenue (5% or more) in supply chain and inventory management,” yet only 6% of companies report using genAI regularly in that market. If it's so effective at increasing revenue, why would more companies adopt genAI?
Again, this is not to suggest that genAI, and AI more broadly, won't have a big impact; rather, it shows that we are still early in the adoption cycle.
Let's get started. Let's break things.
I think one of the main reasons why sales and marketing are the areas where genAI is most used within companies, according to the McKinsey study, is the recognition that these are areas where companies can “get it wrong.” I'm not saying these areas aren't important; I'd rather have someone with a law degree hallucinate on the initial version of their marketing copy than on their P&L. According to McKinsey, the companies that perform best with genAI tend to be the ones that “experienced every negative impact of genAI that we asked about, from cybersecurity and personal privacy to explainability and IP infringement.” They got burned by genAI and are learning from that experience. It's best to learn the ropes from relatively low-risk activities behind your firewall.
Because these high performers understand how to manage risk on the rough edges, they run more genAI workloads than their peers (using genAI in three functions on average, compared to an average of two for less experienced companies). They also have more advanced risk mitigation strategies and are “more than three times more likely to be using genAI than other companies,” according to McKinsey. [more advanced] “They are involved in a wide range of activities, from processing accounting documents and risk assessments to R&D testing, pricing and promotions.” They also face data challenges, with 70% of high performers citing data issues such as lack of understanding of data governance processes and training data.
If you're not willing to experiment and risk breaking things, you'll never encounter these problems and never learn from them.
Going back to Elastic's 1,000 customers paying to build genAI applications, this is great news for Elastic and for the industry, regardless of the short-term financial impact. As company executives stated, while customers are still early in their adoption cycle, genAI will be “a big growth driver for us in the long term.” The way every company moves from early trials to enterprise transformation is to start small, disrupt a few things, and gain the experience and confidence to do bigger things with genAI.
Copyright © 2024 IDG Communications, Inc.
