Real Good AI’s Rating Scale

Let’s talk about AI. Not just yes or no but a more nuanced conversation.

AI Rating Scale*

2026

*draft as of 12/17/2025

  • No intentional AI usage in production. Example: incidental google searches with included AI summaries are ok if not referenced (since these things are impossible to turn off!).

  • Single components of the work like individual words or grammar assistance. Specialized tools or filters (that are not full LLMs/image/video generation) that apply an after effect, filter, or change only a few words or details from your work. This can include the creation of non-central automated accessibility features like captions or transcripts.

  • AI idea development, organization, structuring or outlining, where no final text or images are AI generated. Use for directions, feedback, or other tasks only when human in the loop prevents copy/pasting directly from AI output to the final product.

  • AI generation of isolated phrases or other building block components (even if edited or traced). Think ship of Theseus: if some of the smaller components to a video, image, written work were at any point copy and pasted from AI, it is still influenced by the AI work and should be marked as this level. AI generation likely does not change the overall content or meaning of the work and can include tasks such as cleaning up small components of background with AI generative fill or AI enhancement of voice.

  • AI generation of paragraphs, major components, or major ideas contributed by AI (even if edited or traced). This is more than incidental use, where the content of the work would have been different if AI was not involved. Authorship or otherwise credit would be required if a human had made similar contributions.

  • AI generation of the large part or whole work (even if edited or traced). No matter how complex the prompt is (including supplying reference materials in prompting), if the work is copy/pasted with only minimal human intervention. Also anytime there is active AI interaction or content reaching the user with no human oversight.

Download the 2026 AI Rating Scale Assets

This rating scale was created to help facilitate the self-reported disclosure of the use (or lack thereof) of AI in products, media, education, and other venues. This can be used to asses risk and allows organizations or individuals to set nuanced boundaries in AI use.

Technology is constantly changing so this system will need to be updated and clarified when used in different applications or when new tools launch.

This needs to exist NOW before it is too late!!!

Dr. Mandy introduces the rating system.

How Do I Use the AI Rating Scale?

  1. Read the basic criteria for the levels and decide what you want to use on your project.

  2. Continuously check in with the AI Rating Questionnaire to make sure you adhere to the proper level of use

  3. Disclose your use on publication with the associated graphic above representing your determined rating on your product

  4. Gather feedback from the community for future AI use preferences!

AI Rating Questionnaire

This series of questions are designed to help you rate your work!

~~2026 Version Under Construction~~

Do you like this scale or its possibilities?

Please consider donating to support its continued operation!

Donate

Still have questions about using the AI rating scale? Let us help!

contact@realgoodai.org