Analyze generative AI use in Knowledge Authoring

If you are using generative AI in your knowledge creation process it will now be possible to analyze how it is used by your authors.  Many subject areas have been updated to include new or updated attributes as described below.

Example graph showing versions created using generative AI vs. those that were not.

This will allow you to monitor trends in the use of these features as well as try to understand what impact they might be having on your end user's experience.

Steps to Enable

You don't need to do anything to enable this feature.

Tips And Considerations

Understanding the data

  • Attribute Employed Generative AI - At this time, generative AI only comes into play in the process of creating articles, meaning that v0.1 in a locale is the only one for which this attribute will ever by "Y" currently.  In the future, as more generative AI features are added, this could be Y on other versions.  If you are interested in seeing something like what percentage of articles employed generative AI, you may want to filter down to "Article First Version = Y".
  • Attribute Created by File - This attribute tells you whether or not the article was created by generating it from a file.  This attribute will be "Y" for all versions of an article in the locale in which the generation occurred.
  • Attribute Created by Service Request - This attribute tells you whether or not the article was created using content from an SR.  This attribute will be "Y" for all versions of an article in the locale in which this occurred.  Note that this will be Y whether the article was created using an Action Plan or using generative AI.  In order to determine whether or not it was generation (assuming you are using both techniques) you would need to analyze the first version of the article and the value of Employed Generative AI.
  • Subject with Article - Latest Version - If the subject area you are working with includes this folder, note that the moment you pull a fact into your analysis then ONLY the latest version of the article will be used.  In other words, if you are building a report against the Article Rating Real Time subject area, you can pull in the Article ID, Version, and Employed Generative AI to display a list of all article versions that exist, but the moment you pull in Average Rating then you will only be seeing the latest versions of those articles and Employed Generative AI will equal "N" because it is not possible to rate draft articles.  If you are not using Action Plans to create articles from SRs, then you could use the Created by File and Created by Service Request attributes in this subject area to understand the ratings of articles created using generative AI vs. those that were not.