Dive Brief:
- Businesses are pushing forward on AI initiatives, but most are lacking the technical maturity needed to scale projects, according to a F5 report published Wednesday, which surveyed more than 700 IT decision-makers across industries worldwide.
- Almost two-thirds of IT leaders are worried about compute costs from AI model development, training and inference, according to the survey. The security of models was the second most common concern.
- Adoption roadblocks lie within enterprises’ technology data foundation as well. Nearly the same percentage of IT leaders who are implementing AI (75%) report significant data quality issues (72%).
Dive Insight:
Enterprises are hastily adding AI to their tech stacks, but CIOs who prioritize speed risk losing the full potential of these projects.
“Many enterprises, in their eagerness to harness AI, overlook the need for a solid foundation,” Kunal Anand, EVP and CTO at F5, said in a statement. “This oversight not only diminishes the effectiveness of their AI solutions but also exposes them to a multitude of security threats.”
IT decision makers expect to spend an average of 44% more on security over the next few years as AI deployments scale, according to the report.
Managing risks and resources is vital to AI implementation. JPMorgan President and COO Daniel Pinto said during a May investor call that its AI use cases, valued between $1 billion and $1.5 billion, are meaningless without efficient risk mitigation and resource allocation.
Plus, organizations are looking for deployment opportunities that are innovative and add a competitive advantage, requiring more than just turning access to a model on. Colgate-Palmolive CEO Noel Wallace said the company is not going to get distracted in its pursuit by small wins and hype, during a conference last week.
While AI use in the workplace is rising, the emerging examples focus on less strategic functions, according to the F5 report. Employee productivity tools are the most common use case, and automation remains a priority.
