Light

AI was used for limited support tasks that did not substantially shape the article’s core ideas, structure, analysis, or final claims.

​

Examples: grammar suggestions, title brainstorming, formatting help, search-term generation, minor copyediting.

​

​

​

 

 

Moderate

AI contributed to idea development, structure, or revision in meaningful ways, but the core argument, analysis, and written form remained primarily human-developed and human-directed.

 

Examples: brainstorming sections, outlining, revising prose, testing phrasing, helping organize an already-developed manuscript.

​

​

​

​​

 

Substantial

AI played a significant role in developing, extending, revising, or pressure-testing the work. The human author remained clearly in charge of judgment, evidence, and final decisions, but the article’s final form was materially shaped by sustained AI involvement.

 

Examples: theory development, major drafting/revision assistance, codebook generation, cross-model critique, figure development, repeated reconstruction of arguments.

​

​

​

 

 

Central

AI collaboration was not only substantial, but integral to the article’s method, evidence, object of study, or authorship structure. The work could not be accurately described without foregrounding the human–AI collaboration itself.

 

Examples: the article studies AI dialogue directly, depends on extended AI interaction as part of the dataset or method, or presents AI collaboration as a central constitutive feature of the work.