Deliberate AI Use
The always-on rules (explain choices, mark red-zone code, ask before delegating fully) are in deliberate-ai-use.instructions.md. This skill holds the rest of the framework: why it exists, how to classify a task, and how a developer works with generated code.
Why
Research shows that how you use AI matters more than whether you use it. In Anthropic's 2026 study, developers using AI assistance averaged 50 % on a knowledge quiz, against 67 % for those who coded by hand. Those who delegated everything scored under 40 %. Those who asked questions to understand the code scored 65 % or more (Anthropic).
The framework makes sure AI tools strengthen developers' skills instead of weakening them.
Sources:
- Anthropic: How AI assistance impacts coding skills (2026)
- INNOQ: AI Coding Patterns Through Cognitive Load Theory (2026)
- METR: AI experienced OS dev study (2025)
- Stray et al.: Developer Productivity With and Without GitHub Copilot (HICSS-59, 2026), a study of Nav IT
- MIT/Microsoft: The Effects of Generative AI on High-Skilled Work (2025)
- Nav developer survey 2026: 59 % are worried about losing skills
Green and red zone
Classify a task before using AI on it.
🟢 Green zone: suited to AI
Tasks where AI adds the most value without weakening understanding:
- Boilerplate and repetitive code (Nais manifest, CRUD endpoints, Dockerfile)
- Technology the developer already knows well
- Configuration and infrastructure
- Refactoring with a known goal (rename, extract, move)
- Test data and fixtures
🔴 Red zone: code by hand first
Tasks where coding by hand builds critical skill:
- Debugging: troubleshooting is the strongest learning mechanism
- New concepts: technology the developer has not used before (learn first, generate afterwards)
- Core business logic: business rules, calculations, state machines
- Security-critical code: authentication, authorization, input validation
- Architecture decisions: system design, data models, API contracts
Three-attempt rule: try to solve the problem yourself, with at least three attempts (approaches), before asking AI for help. Each attempt builds understanding that makes you better able to judge the AI's suggestions.
Experience level: junior developers should keep more in the red zone. Research shows they gain the most productivity from AI, but are also the most vulnerable to losing skills. Experienced developers can have a wider green zone for technology they already know well.
Generate-then-understand
When AI generates code, do not just accept it. Use the generate-then-understand pattern:
- Generate: let the AI write the code
- Understand: ask why the code is written the way it is
- Verify: check that you can explain every part yourself
- Adapt: make deliberate changes, not just copy-paste
Good follow-up questions
Offer these to the developer as written, in Norwegian:
textHvorfor valgte du denne tilnærmingen fremfor alternativene? Hva kan gå galt med denne koden? Hvilke edge cases dekker den ikke? Forklar tradeoffene i denne designbeslutningen
