It’s clear that there’s a lot of interest in all things involving artificial intelligence (AI) across the enterprise, but it’s also clear that the risk in return on investment (ROI) is among the highest any organization is going to make. Most data scientist teams are lucky to deploy two AI models a year in a production application. The trouble is that the business conditions on which the AI models are predicated are subject to change. Never has that been more apparent than in the wake of the COVID-19 pandemic. Entire processes have been transformed in ways that rendered obsolete much of the data used to train AI models using machine learning (ML) algorithms.
More than 40% of large company IT directors said the pandemic affected at least half of all their AI/ML projects from a priority, staffing or funding standpoint, according to a recent survey by AI tools provider Algorithma.
Over half the respondents (54%) said AI/ML projects prior to the pandemic were focused on financial analysis and consumer insight. In the wake of the pandemic survey respondents said more of their AI focus is now on cost optimization (59%) and customer experience (58%).
The survey also notes respondents were spending at least $1 million annually on AI/ML prior to the pandemic, with 50% saying they are planning to spend more than that going forward.
AI becomes higher priority
A total of 43% of the respondents also noted AI/ML matters much more than they thought now as a result of the pandemic, with 23% reporting they realize now that AI/ML should have been their highest priority IT initiatives all along. Overall, nearly two-thirds (65%) said that AI/ML projects were at or near the top of their priority list before the pandemic, with 33% now saying these applications are now higher on their list.
Many organizations now realize that at a time when they simply may not have enough people on hand to manage processes the way they once did there’s going to be more of an effort to reduce costs, says Algorithmia CEO Diego Oppenheimer.
“The downturn creates a forcing function,” says Oppenheimer. “Organizations need to get AI models into production.”
However, while the survey makes it clear there are still a lot of investments being made in AI, it also suggests roughly a third of respondents have not made AI a similar priority. In addition, it’s also apparent that half of respondents won’t be increasing investments in AI.
It’s not clear to what degree the level of spending on AI reflects the nature of the project or is simply a response to macroeconomic conditions affecting a specific vertical industry.
Regardless of motivation, when it comes to spending it doesn’t appear building and deploying AI models is going to get that much easier anytime soon despite advances in AI operations (AIOps). In fact, it may be more important to pick the right use cases for AI than ever at a time when the tolerance for experimentation within most organizations has in recent memory never been lower.