Update README.md

README.md

@@ -96,13 +96,15 @@ Generative AI safety requires ongoing expertise and collaboration, and we believ
 
 We invite the community to contribute to the ongoing development of **LISAT-7B**, and we provide an open-source toolset to help developers assess and deploy the model in a safe and responsible manner.
 
-### Ethical Considerations and Limitations
+### Ethical Considerations
 
-**LISAT-7B** has been designed to be helpful and inclusive in remote-sensing applications. It is meant to serve a wide range of use cases, from environmental monitoring to urban planning, while addressing user needs without unnecessary judgment or normativity. We are committed to ensuring that the model reflects diverse perspectives, especially when it comes to interpreting sensitive data.
+This paper presents advancements in reasoning segmentation for remote sensing tasks. **LISAT-7B** is a method that is able to reason over arbitrary remote sensing images and output both explanations and segmentation masks for objects of interest. These kinds of workflows are extremely common across multiple fields. For example, disaster management personnel may want to know which roads leading to an airport are undamaged, and why. **LISAT-7B** is the first such model that can simultaneously answer both components of such questions.
+
+Broadly, **LISAT-7B** has impacts in numerous domains such as environmental monitoring, urban planning, and search and rescue. However, one of the biggest uses of satellite is surveillance. Our work would also aid in intelligence use cases whether they be conducted by friendly governments or malicious actors. We offset this use case by basing GRES primarily on the xView series of datasets which are explicitly created with AI for good purposes in mind. This means that any intelligence use of **LISAT-7B** would require further dataset curation and training.
+
+We encourage responsible deployment and continued discourse on the implications of geospatial AI in real-world applications.
 
-However, like any technology, **LISAT-7B** comes with inherent risks. Testing conducted thus far cannot cover every possible scenario. As a result, the model may sometimes generate inaccurate, biased, or otherwise inappropriate outputs, particularly in complex geospatial or ambiguous settings. Developers using **LISAT-7B** for their applications should perform extensive safety testing and fine-tuning tailored to their specific use case to ensure responsible and ethical deployment.
 
-Please, use **LISAT-7B** responsibly.
 
 ## Citation