AI Transparency

Last updated: 03-03-2026

This blog uses artificial intelligence (AI) tools in a limited, deliberate, and supervised way. This page explains how AI is used, how it is disclosed, where it is not used, and who remains responsible for everything published here.

This policy is not promotional. It is descriptive.

TL;DR

I use AI as a supervised tool for brainstorming, structuring, language polishing, translation, and software development. All content published here is written and validated by me. Each post carries a structured AI usage indicator — a level from 0 to 3 for text, and 0 to 2 for code — so readers always know the degree of AI involvement. AI never replaces authorship, responsibility, judgment, or moral positioning.

1. Core principle

AI is an assistive tool, not an author, not a decision-maker, and not a moral agent.

All published content on this blog is written, curated, and validated by me. When AI is involved, it is used to support thinking and execution, never to replace responsibility, authorship, or judgment.

2. AI usage levels — text content

Each post may carry a text AI usage level. These levels reflect how much AI contributed to the written content.

Level 0 — No AI involvement

The content was written, edited, structured, and proofread entirely without AI tools.

Examples: personal reflections, short political analysis, advocacy posts written end-to-end manually.

Disclosure label: AI: None

Level 1 — Editorial assistance

AI was used only for surface-level improvements. The substance, structure, argument, and narrative are fully human.

Typical uses: grammar correction, title suggestions, minor rephrasing, tone refinement, translation, formatting to Markdown.

Authorship: 100% human-led. Disclosure label: AI: Editorial assistance only

Level 2 — Co-drafting assistance

AI helped expand draft ideas, restructure content, or generate paragraphs based on my outline. The intellectual direction, thesis, and validation remain mine.

Typical uses: turning bullet points into paragraphs, expanding a draft into a structured essay, generating alternative arguments that I then refine, drafting tutorial text based on my actual codebase, summarizing material I already read.

Authorship: Human-directed, AI-accelerated. Disclosure label: AI: Co-drafting assistance

Level 3 — AI-generated, human reviewed

The majority of the textual body was generated by AI based on a prompt. I acted primarily as reviewer, editor, fact-checker, and curator.

Typical uses: full tutorial generation from a codebase, long technical explanations mostly AI-written, generated content where I mainly verify and adjust.

Authorship: AI-generated, human validated. Disclosure label: AI: AI-generated (human reviewed)

3. AI usage levels — code content

Code has its own axis because the authorship question is different for implementation than for ideas.

Code Level 0 — Human-written

All code was written manually without AI generation tools.

Disclosure label: Code: Human-written

Code Level 1 — AI-assisted

AI helped with boilerplate, debugging, refactoring suggestions, or generated portions of the code. The architecture, logic, and decisions are human-led.

Disclosure label: Code: AI-assisted

Code Level 2 — Primarily AI-generated

Most or all code was generated by AI — prototypes, scripts, infrastructure templates, or refactored code. All AI-generated code is reviewed, understood, and validated before use or publication.

Disclosure label: Code: Primarily AI-generated

4. How disclosure works

Each post includes structured AI usage metadata:

• Text level (0–3) — how much AI contributed to the written content
• Code level (0–2) — how much AI contributed to code examples
• Tools used — which AI tools were involved (e.g. Claude, ChatGPT, Copilot)
• Description — optional short human-readable note

This metadata is machine-readable, displayed on the frontend, and allows readers to filter or assess posts by AI involvement. The goal is precision — not vague “AI was used” disclaimers.

If a post does not contain this metadata information, its likely written by me.

5. What AI is NOT used for

Unless explicitly stated otherwise:

• AI does not generate full blog posts autonomously
• AI does not produce personal testimony or lived experience
• AI does not define political positioning or moral judgment
• AI does not introduce factual claims without independent verification
• AI is never cited as a source
• AI does not replace critical analysis or accountability

The voice, intent, and responsibility of each text remain mine.

6. Authorship and responsibility

All blog posts are fully written by me, or — when AI materially contributes — directed, reviewed, and validated by me.

I take full responsibility for:

• Factual accuracy
• Interpretations and opinions expressed
• Errors, omissions, or ambiguities

AI assistance does not transfer or dilute authorship or accountability. The AI usage level clarifies how the tool was used, not who is responsible. That is always me.

7. Accuracy, sources, and verification

To limit the well-known risks of AI systems (hallucinations, confident errors, false citations):

• AI is never cited as a source
• AI-suggested facts are systematically verified
• AI is used to rephrase or clarify, not to introduce new information

If an error is identified, it is corrected regardless of whether AI assistance was involved.

8. AI use in software development

AI tools (notably Claude Code) are used in software work for:

• Boilerplate generation
• Debugging assistance
• Architecture discussions
• Refactoring suggestions
• Documentation drafting
• Security review support

All AI-generated code is:

• Reviewed manually
• Understood before use
• Validated & tested before deployment

AI is treated as an accelerator, not an authority. Code provenance, licensing, and data exposure risks are actively considered.

Code projects published on this blog carry a Code level (0–2) to indicate the degree of AI involvement.

9. Professional use of AI

AI is also used in a professional context, under strict constraints:

• No sensitive data is shared
• No personal data is disclosed
• No internal documents are provided
• No credentials or secrets are exposed

Employers, systems, and internal processes are never named or described. Confidentiality is non-negotiable.

10. Positioning and limits

This blog does not present AI as neutral, ethical by default, or legally unproblematic.

AI systems are understood as:

• Invasive by nature
• Trained on opaque data sources
• Embedded in economic and power structures

Using AI pragmatically does not imply endorsement of the broader AI ecosystem.

This policy exists precisely to draw boundaries, not to normalize uncritical use.

11. Audience and intent

This policy is written for:

• Readers seeking transparency
• Critics and journalists evaluating credibility
• Potential collaborators
• Myself, to apply transparency in practice

If my use of AI changes materially, this page will be updated accordingly.

Transparency is not a claim. It is a practice.