About
A tool for transparent research
AI Usage Label Generator
Nutrition-facts–style labels for disclosing how AI was used in scholarly work.
As generative AI becomes part of everyday research and writing, disclosure statements remain inconsistent, vague, and difficult to compare. The AI Usage Label Generator adapts a familiar format — the nutrition facts panel — to make AI disclosure structured, granular, and legible at a glance. Instead of a single sentence buried in a methods section, contributors produce a clear label that reports AI involvement stage by stage, alongside the verification steps taken to check that work.
What it is
The tool is a free, browser-based generator that turns a short, guided reflection about your workflow into a standardized visual label. It runs entirely in your browser, requires no account, and stores nothing on a server. The label can be embedded in a manuscript, appended to a submission, shared alongside a dataset, or handed in with a course assignment.
How it works
- Choose a workflow template, or design your own set of research stages.
- For each stage, indicate the degree of AI contribution on a graded scale.
- Note the specific tools and models used, and record the checks you performed.
- Preview the label in real time, then export or share it.
Features
Stage-by-stage disclosure
Report AI involvement across the phases of a project — ideation, literature review, methodology, data collection, analysis, writing, visualization, and more — rather than as a single undifferentiated statement.
A graded scale, not a yes/no
Contribution is recorded on a five-point scale (0%, 25%, 50%, 75%, 100%), so partial assistance is represented as partial. Stage and overall percentages are computed automatically, weighted by the epistemic significance of each stage.
Workflow templates and a custom designer
Preconfigured starting points cover common project types — literature reviews, experimental studies, data-analysis projects, theoretical papers, and minimal-AI work — plus book and grant proposals. A workflow designer lets you add, reorder, and weight your own stages, then save, import, and export them for reuse.
Verification and integrity checks
Alongside usage, the label records the human oversight applied: human verification, citation checking, plagiarism screening, data-leakage and ethics/IRB considerations, and attention to hallucination and missing-literature risks.
Live preview
The label updates instantly as entries are made, so what you review is exactly what you share.
Portable outputs
Export as PNG or SVG, copy to the clipboard, save the structured data as JSON, or attach a QR code linking to a DOI or full disclosure record.
Design philosophy
The tool rests on a simple premise: AI functions less like a hammer and more like a cognitive prosthetic — it extends our linguistic reach, augments memory, and distributes our thinking across a human–machine assemblage. The text that results is neither purely the author's nor purely the machine's, and that blurring is often invisible even to the person doing the writing. Disclosure, in this view, is not a confession of inadequacy but an acknowledgment of a new mode of working.
Why a nutrition label?
The nutrition facts format was chosen deliberately. Like a food label, the AI Usage label standardizes disclosure in a recognizable form, quantifies something previously vague, itemizes its components, and flags potential risks — all so that readers can make their own informed judgments. The metaphor carries a message too: just as processed food is not inherently bad but consumers deserve to know what is in it, AI-assisted work is not inherently inferior, but readers deserve to know how it was made.
Disclosure as self-knowledge
Filling out a label is designed to be a reflective practice, not bureaucratic box-checking. To complete one, an author must reconstruct their workflow, classify each stage by AI involvement, and confront the gap between what they felt they did and what actually happened. The act of articulating AI's role is what forces you to know it.
Not all assistance is equal
Using AI to brainstorm search keywords is not the same as using it to write a discussion section. The label encodes this through customizable epistemic weights, so that contribution at more consequential stages counts for more. The aim is not false precision but structured reflection.
Any disclosure is better than none. Perfect recall is impossible; demanding it only discourages disclosure altogether. An approximate, honest label still conveys crucial information; the practice of making one produces value in itself.
Non-judgmental by design
The label discloses; it does not condemn. It uses neutral language (“contribution,” not “contamination”), applies quality controls equally relevant to all research, and assigns no grade or score to AI involvement. The goal is informed evaluation, never prohibition.
Who it is for
Researchers, instructors, and graduate and undergraduate students— anyone who wants a clear, consistent way to communicate AI use — and to support academic integrity, reproducibility, and informed peer review. By giving AI use a shared visual grammar, the label helps authors disclose honestly and helps readers understand at a glance what they are evaluating.
Try it: ailabel.netlify.app — free, no account required.
