AI speeds up coding and creates new bottlenecks, GitLab survey finds

It’s no secret that artificial intelligence is transforming how software developers do their jobs, but the impact of AI goes much deeper than how developers write the code.

GitLab’s annual DevSecOps report reveals what the company is calling the "AI Paradox," in which AI certainly accelerates the coding work.

But AI is also creating new bottlenecks in testing, security, compliance and team coordination that cost organizations nearly a full workday per engineer each week.

For example, the report’s survey found that DevSecOps professionals lose 7 hours a week. This is because of inefficient processes such as poor communications, limited knowledge sharing and different tools used across teams.

GitLab's survey of around 3,200 DevSecOps professionals found that 60% of their teams used more than five tools for software development.

Agentic AI will be successful when implemented in a platform engineering approach, respondents told GitLab.

Roles in the software development cycle are changing, which is driving the pressure to upskill. Coding is getting easier but there is a need for software engineers, 76% of respondents said.

Eighty-seven percent believe that software engineers who adopt AI are future-proofing their careers, while 83% said that AI will significantly change their roles over the next five years.

Despite widespread AI adoption, respondents to the GitLab survey said that human oversight remains essential.

While 97% of respondents are using or plan to use AI in software development, only 37% would trust AI to handle daily work tasks without human review.

AI also creates compliance complexity that demand human expertise today, but many of the respondents said that compliance will be built directly into code.

“Organizations need a new framework to match the speed of software development in the age of AI, one that provides intelligent orchestration across the entire software lifecycle while addressing the interconnected requirements of AI orchestration, governance, and compliance that individual point tools simply cannot solve,” said Manav Khurana, chief product and marketing officer for GitLab.

The entire report can be found here.