Who adopts — and who holds back
Adoption varies by experience, role, company size, and geography. Senior developers are more cautious, small companies favor different tools than enterprises, and policy still blocks 1 in 10 teams entirely.
Daily AI Use by Career Stage
Stack Overflow 2025
Experienced developers report higher "highly distrust" rates — consistent with code-review responsibility and accountability.
AI-Generated Code % by Role
HackerRank 2025 — which roles lean hardest on AI output
Security, cloud, and data engineers lead AI code generation. Roles with more boilerplate and infrastructure code show higher AI reliance.
Tool Preference by Task
HackerRank 2025 — developers choose different tools for different jobs
Copilot and Claude favored for refactoring; Cursor stands out for testing, integration, and deployment tasks. ChatGPT dominates general coding Q&A across all tasks.
Enterprise Policy Posture
JetBrains Developer Ecosystem 2024
~80% allow AI tools to some extent; 11% prohibit third-party cloud AI outright. Regulated industries are slowest to adopt.
How Much Code Does AI Write?
HackerRank 2025 — depth of reliance on AI-generated code
"Almost always" AI users report 48% of their code is AI-generated and 37% say AI writes the majority. Heavy users generate ~2x more AI code than the average.
Tool Choice by Company Size
Pragmatic Engineer 2026 (n=906) — procurement dynamics drive divergence
Small companies favor Claude Code and Cursor (developer choice). Large enterprises default to Copilot (procurement, compliance, SSO). Sample is self-selected experienced engineers — treat as signal, not population data.