Playing with the wording a bit.

Also not sure I understand the "hidden" .....

app.py

@@ -488,10 +488,12 @@ with gr.Blocks(css="#visualization-window {flex-direction: row-reverse;}") as de
     with gr.Accordion("About this demo 👀"):
         gr.Markdown("## What's in your dataset?")
         gr.Markdown("""
-        Addressing fairness and bias in machine learning models is [more important than ever](https://www.vice.com/en/article/bvm35w/this-tool-lets-anyone-see-the-bias-in-ai-image-generators)! One contributor to disparate performance is an imbalanced training dataset. Social biases will affect dataset curation, but that's difficult to investigate without specific features to disaggregate for subpopulations. However, even though a dataset might not have these as _explicit_ features, these characteristics may still be hidden in the data.
+        Addressing fairness and bias in machine learning models is [more important than ever](https://www.vice.com/en/article/bvm35w/this-tool-lets-anyone-see-the-bias-in-ai-image-generators)! 
+        One form of fairness is equal performance across different groups or features. 
+        To measure this, evaluation datasets must be disaggregated across the different groups of interest.
         """)
 
-        gr.Markdown("The `disaggregators` library ([GitHub](https://github.com/huggingface/disaggregators)) provides an interface and a collection of modules to help you disaggregate datasets by their hidden features. Click through each of the tabs below to see it in action!")
+        gr.Markdown("The `disaggregators` library ([GitHub](https://github.com/huggingface/disaggregators)) provides an interface and a collection of modules to help you disaggregate datasets by different groups. Click through each of the tabs below to see it in action!")
         gr.Markdown("""
         After tinkering with the demo, you can install 🤗 Disaggregators with:
         ```bash