Report by Black Duck
The State of AI-Powered Software Development
Key Findings
48% of development teams experience bottlenecks in code rework related to AI-generated code.
Teams with full governance for AI coding assistants in place are 55% more likely to report a major improvement in efficiency.
64% of development teams express moderate or extreme concern about AI coding assistants introducing security defects or vulnerabilities.
Developers reclaim an average of eight hours per week when using AI coding assistants.
Nearly 90% of development teams encounter issues with AI-generated code.
86% of development professionals believe an AI agent or model should evaluate AI-generated code.
56% of development professionals prefer a dedicated AI security agent separate from the code-generation tool to evaluate AI-generated code.
30% of development professionals believe the same AI model that generated the code should also review it for security issues.
84% of developers prefer to keep a human in the loop via pull requests or real-time IDE suggestions when using AI-assisted development.
AI coding assistants have 97% adoption among enterprise development teams.
97% of development teams have adopted AI coding assistants.
92% of development teams report improved productivity and release velocity when using AI coding assistants.
30% of development teams have full governance in place for AI coding assistant adoption and oversight.
51% of development teams experience bottlenecks in security testing related to AI-generated code.
53% of development teams have grown total code volume by over 25%.
52% of development teams experience bottlenecks in manual review related to AI-generated code.
68% of developers say it is extremely important to have a clear, automated system for tracking AI-generated code and measuring its impact for debugging, security, and accountability.
58% of development teams cite a major improvement in productivity and release velocity from AI coding assistants.
Developers will spend 29% more time reviewing and validating AI-generated code, 29% more time on complex architecture and system design, and 23% more time on security verification and risk management.