Mike Vizard summarizes the EMA/Perforce survey findings, exploring how AI coding tools are shifting focus in DevOps workflows, with increased time spent on review and new challenges around security and code quality.

Impact of AI Coding Tools on DevOps Workflows

Survey Overview

A survey conducted by Enterprise Management Associates (EMA) on behalf of Perforce, polling 206 IT leaders and executives, reveals the significant influence AI coding tools have on modern DevOps workflows.

  • Review > Writing: 57% of developers are now spending more time on reviewing code, standards, and quality oversight than actual coding.
  • Security & Compliance: 53% of respondents take on greater security, policy, and compliance responsibilities. Validating AI-generated code is a priority for 52%.
  • Tool Limitations: Jake Hookom (Perforce) notes teams invest extra effort to address limitations of AI coding tools, such as code verbosity, performance impact, and technical debt.

Key Findings

  • Security Risks Top Concern: 62% identify security and privacy as the biggest challenge when using AI tools for coding.
  • 52% fear introducing vulnerabilities or code defects.
  • Two-thirds worry about reliance on AI tools, with 61% citing “blind faith” in AI-generated results.
  • 57% report negative or neutral experiences with inconsistent code quality/testing results from AI tools.

Organizational Responses

  • Investment Justifications: AI tool investments are tracked via improvements in code quality, defect reduction (70%), and developer productivity (62%).
  • Developer Perception: Developer satisfaction (62%), faster time to market (49%), smooth onboarding of junior developers (43%), and better test coverage (56%) are notable positives.
  • Challenges: Risks include poor/insecure code (54%), integration difficulties (45%), limited architecture/design controls (44%), ownership concerns (33%).

Adoption and Workflow Implications

  • 51% report active use of ‘vibe coding tools’; only 3% feel these will fundamentally reshape development workflows.
  • Benefits: Productivity enhancement (38%), lower barriers for new/returning developers (33%).
  • Areas for improvement: Real-time vulnerability detection (55%), automated test generation (53%), pipeline orchestration (46%), AI-assisted performance testing/environment simulation (46%).

Blocking AI Coding Tools

  • When AI tools are blocked, top reasons are security (45%), intellectual property protection (32%), vendor conflicts (25%).
  • Most organizations are supportive or somewhat supportive of using non-approved tools despite concerns.

Conclusion

While adoption is early, the discussion has shifted from “if” to “to what extent” organizations use AI in software development. The interplay between benefits and risks is driving a re-evaluation of DevOps workflows, with increased emphasis on review, validation, and process adaptation.

References:

Key Takeaways

  • AI coding tools are a double-edged sword: boosting productivity, but raising significant security and workflow challenges.
  • Organizations are investing in adaptation, but full process re-engineering remains uncommon.
  • Validation, review, and security overlays are vital when deploying AI-powered coding solutions in DevOps pipelines.

This post appeared first on “DevOps Blog”. Read the entire article here