This policy sets the standard for safeguarding high-quality professional work when using AI as a tool to assist in completing that work. It applies to any professional work, technical or non-technical, across any field or job type. We explicitly disavow and prohibit the production of "AI slop": any code, documentation, analysis, writing, or other professional work product generated without active human design, rigorous direction, and line-by-line verification before it is submitted or delivered as professional work.
AI and its underlying LLMs are not autonomous professionals that a user can simply command to design, create, and produce accurate, trustworthy work on their behalf. AI is a biased, non-deterministic, pattern-matching model trained on large datasets.
AI must be treated as just another instrument in your technical toolbox. However, unlike deterministic, context-free tools, AI is biased and non-deterministic, and if used inappropriately in a professional setting, it can have devastating consequences.
Example scenarios:
- Commercial airplane software running on AI slop code.
- Doctors relying on AI slop software to assist with diagnostics.
- Trusted media reporting AI slop analysis as fact.
AI core operational risks:
- Non-Deterministic: The exact same input will not consistently yield the exact same output.
- Bias and Hallucination: Models produce plausible-sounding answers even when completely wrong, and often default to telling you what you want to hear rather than what is accurate. The model owners can also build in biases and guardrails that further shape responses.
- Pattern Matching, Not Reasoning: AI mimics human communication patterns; it does not possess genuine domain understanding, intent, or critical thinking.
- Cognitive Law: All ideation, architectural design, problem definition, and strategic decisions MUST originate entirely from the human professional throughout the process.
- No Delegated Execution: You are strictly prohibited from acting as a passive prompter who hands off end-to-end thinking, development, or analysis to an AI.
Rule II: Low-Consequence, Highly Generative Use for Staging, Demo, Testing, and Inconsequential Work
Fully generative AI output may be permitted in non-critical scenarios where a mistake carries minimal to no negative consequence, for example: creating a temporary video thumbnail, drafting placeholder graphics, generating mock test data, flagging patterns where a false positive is a non-issue, or mocking a quick proof of concept.
- Agentic AI may be used in this scenario.
- All fully AI-generated work must be openly declared as such.
- Whether a minor part of production work may be fully AI-generated depends on the field. A YouTube video thumbnail may be reasonable to fully generate, but an anatomy diagram used to teach medical students should not be, since accuracy is critical to that field.
AI may be leveraged to eliminate mundane, high-friction work that is easy for a human to review. In this case, AI should be used to assist with small chunks of work, each reviewed individually, rather than one large, unreviewable batch. Examples include:
- Draft Refinement: Polishing grammar, formatting, and tone on human-authored manuals or documentation.
- Metadata & Overhead: Generating standardized pull request templates, tags, warnings, or structural boilerplate.
- Repetitive Implementation: Expanding simple, repetitive patterns across known identifiers or schemas (e.g., writing code to map enum to strings).
- No Fully Generated Work Submitted as Professional Work: AI must never be used to generate an entire codebase or complex end-to-end project intended for production use.
- Mandatory Line-by-Line Audits: Any AI-assisted code, query, PR description, or document must be thoroughly inspected and verified by the human author before commit or publication.
- The Accountability Standard: You are 100% accountable for all work you produce, especially when you use AI to assist you. If you cannot explain, defend, or manually debug a piece of the work, it is AI slop and must not be submitted.
- Mandatory Disclosure: You must openly declare all AI use, whether full or partial. For example:
- "Part of this work was generated using an artificial intelligence tool or product."
- "This entire product was generated using an artificial intelligence tool or product. It has not been independently verified for reliability or accuracy; use at your own risk."
- Optionally, you may provide further detail if you deem it necessary.
Open Professional AI Policy by Alrick Grandison (2026).
To the extent possible under law, Alrick Grandison has waived all copyright and related or neighboring rights to the "Open Professional AI Policy" via the Creative Commons CC0 1.0 Universal Public Domain Dedication.
This work is published from: Canada.
To view a copy of the CC0 legal code, visit:
https://creativecommons.org/publicdomain/zero/1.0/