academic-humanizer README

The project's README from github.com/AIScientists-Dev/academic-humanizer. As one markdown file: README.md.

Academic Humanizer: personalized editing for AI-assisted academic drafts, keeping your voice and every claim, number, and citation intact

license  version  skill  built by

Why we built this

Some of us write a lot of papers and grant proposals, and our team started using AI to help with drafts. The problem is that AI-assisted drafts come out generic and verbose, with "In recent years..." openers, inflated phrasing, and over-long sentences. They also drift from the author's own voice and lose the precision scholarship depends on.

There are tools called "humanizers," but they are built for blogs and marketing. Run one on a paper or an NSF proposal and it flattens the precision along with everything else. The careful wording academic writing depends on is the first thing to go.

So we put together our own. To calibrate it, we had the AI compare its own drafts with our team's accepted papers and funded proposals, and we went through the differences by hand. It is nothing fancy, and it is not about gaming review, defeating detectors, or adding fake novelty. We wanted AI-assisted drafts to read clearly and in the author's own voice, with the numbers, citations, and claims left exactly as written.

Ethics and disclosure

This is an editing aid for clarity and voice, calibrated to an author's own prior accepted work. It does not generate findings, invent data, or change citations, and it is not designed to evade AI-use detection. Using it does not remove your obligation to disclose AI assistance: always follow the disclosure policy of the venue you submit to.

See it work

[!CAUTION] Before (a generic AI draft):

In recent years, continual learning has attracted increasing attention and achieved remarkable success. However, existing methods still face crucial challenges. In this proposal, we propose a novel framework that leverages cutting-edge techniques to delve into these intricate problems, paving the way for a transformative paradigm that will revolutionize the field.

[!TIP] After (clear, in the author's voice, with claims tied to evidence):

Continual learning matters, but today's methods stay empirical and their principles are unclear. That limits reliability and progress. This proposal builds a principled framework on three fronts: adaptation, soft supervision, and cross-domain knowledge. We demonstrate it on autonomous driving and network management.

More before/after passes are in examples/before-after.md: a general example, an NIH Specific Aims page, and a funded NSF CAREER summary.


What it does

Install

git clone https://github.com/AIScientists-Dev/academic-humanizer ~/.claude/skills/academic-humanizer

It is a plain SKILL.md plus examples, so it also runs as a skill or system prompt for Codex and MorphMind. Point your agent at SKILL.md.

Use

/academic-humanizer
[paste a section, or point at main.tex]
# optionally: "match my voice from prior_paper.pdf; target venue: ICLR"

Make it yours

The rules here reflect one group's voice. Fork the repo and adapt them to your own: point it at a few of your past papers, keep the checks that fit your field, and adjust the rest. It is meant to be personalized, not a one-size-fits-all filter.

How it works

Six layers: general AI-tell catalog → academic-specific tells → preserve scholarly conventions → claim↔evidence matching → voice/venue calibration → funding-proposal mode (NSF/NIH structure, first-page primacy, claim↔feasibility). The audit→rewrite loop is defined in SKILL.md.

References

Layer 6 distills the stable structure of NSF and NIH proposals. For current, binding requirements (page limits, formatting, deadlines), consult the source:

Acknowledgments

This skill is the narrower piece: a single-purpose editing pass that sharpens clarity and matches claims to evidence while preserving the author's scholarly voice.

License

MIT.

academic-humanizer (AIScientists-Dev/academic-humanizer)

Searching Database…

What is this?An editing skill that strips typical AI-writing tells from academic papers and NSF/NIH grant proposals. It keeps the scholarly voice, ties claims to evidence, and leaves numbers and citations exactly as written.
Is it a standalone app or something I plug into an agent?It is not a standalone app. It is a plain SKILL.md file plus examples that you load into an AI agent such as Claude Code, Codex, or MorphMind.
How do I start using it?Point your agent at SKILL.md. It can run as a skill or be fed in as a system prompt.
Do I need anything special before starting?You need an AI agent that reads skills or system prompts, such as Claude Code, Codex, or MorphMind. Since it is a plain file, there is nothing to compile or package.
Will it help me get past AI-use detectors?No. The README says it is not designed to evade AI-use detection, and using it does not remove your obligation to disclose AI assistance. Always follow the disclosure policy of the venue you submit to.
Is it only for NSF and NIH?No. It has a general mode for papers and a separate mode for grant proposals. It is meant to be adapted to your own field and past accepted work, not used as a one-size-fits-all filter.
Why did the author build it?The team saw AI drafts come out generic and verbose. Existing humanizer tools are built for blogs and marketing and flatten the precision that academic writing depends on.
How is the editing process organized?Six layers: a general AI-tell catalog, academic-specific tells, preservation of scholarly conventions, claim-to-evidence matching, voice and venue calibration, and funding-proposal mode. The audit-then-rewrite loop is defined in SKILL.md.
Are there examples I can look at?Yes. examples/before-after.md has a general example, an NIH Specific Aims page, and a funded NSF CAREER summary, each showing a draft before and after editing.
What does the grant proposal mode focus on?It keeps the vision that a paper would trim and spends most of its effort on the first pages, since that is what reviewers score.
Size
0.3 MB
Stars
5.4K
Updated
CommitJul 03
Age
4mo
Commits · 90 Days