Last updated:
Transparent AI Assistance
ProdAI Stack covers AI engineering and uses AI-assisted tools in parts of its production workflow. We believe that use should be disclosed clearly.
AI can accelerate research, coding, editing, and implementation work, but it does not replace technical verification or editorial responsibility.
How AI May Be Used
- organizing research and exploring technical questions
- preparing briefs, drafting, restructuring, and editing
- suggesting examples, generating code, and reviewing code
- generating test ideas and refining diagrams or image prompts
- assisting SEO metadata and checking consistency
- identifying gaps and summarizing implementation results
Human Editorial Responsibility
AI-generated output is working material, not authoritative evidence. Final content remains subject to editorial responsibility.
Claims should be checked against official documentation, primary sources, implementation evidence, tests, reproducible experiments, or independent technical review as appropriate.
Code Assistance
AI coding tools may help create or modify reference implementations. A technically verified status must reflect checks that actually ran, such as type checking, linting, tests, deterministic experiments, example execution, and architecture validation.
AI-generated code is not correct merely because it compiles.
Research
AI may help identify subjects, terminology, competing approaches, or potential sources. Important factual claims still require appropriate verification, and AI-generated citations must be checked.
Experiments
AI may help design test cases or harnesses. Published results must come from executed tests or clearly identified measurements—not invented numbers. Synthetic, mocked, simulated, and real-world results must be identified.
Images
AI-generated illustrations are explanatory media. They should not be presented as photographs, measurements, benchmark evidence, or screenshots of systems that were not actually used.
Limitations
AI systems can produce factual errors, outdated information, fabricated citations, insecure code, misleading abstractions, invalid assumptions, and plausible but incorrect explanations.
AI output is therefore not automatically publishable.
Transparency
This policy describes the publication-level workflow; it does not imply every article uses AI in the same way. Additional disclosure may be provided when AI involvement is material to interpreting a result, image, experiment, or implementation.
Privacy and Sensitive Information
Confidential credentials, production secrets, private API keys, passwords, or unnecessary sensitive information should not be intentionally supplied to AI systems.
Our Standard
Is the published material accurate, useful, appropriately verified, and transparent about its evidence?