Primary objective is to inform

**Guidelines:**

Your primary role as a content writer at Softage will be creating a database for Large Language Models (LLMs). You will be provided with a set of queries and the responses given by the AI tool, along with the references used. Your primary responsibility is to edit the response to improve its readability and usability. Analyze the response and the references to identify points of improvement and provide a better, value-adding response. It is also important to understand the context and intent behind the query and try to add a human touch to the answer to train the AI model in natural human language.

#### Editing queries
Provided a user query and the AI's response to it with references, you have to analyze and edit the response to make it more accurate, helpful, and humane. Consider a variety of criteria to edit the response. We aim to add value to the response. Here are some guidelines to follow when editing a query response:
1. Read the responses and citations(if any) thoroughly.
2. Make the response more accurate:
- Check if the information provided in the response is correct as per the latest available data.
- Make sure the response does not have any misleading information.
- Ensure the response does not provide incomplete or partial details that can lead to misinformation.
- Understand the user's intent and context to edit the query. Oftentimes, the AI response can inaccurately guess the user intent.
- Consider any implied information in the user query to make the response more accurate.
3. Make the responses more helpful:
- The language of the response should be easy to understand without overtly complex sentences and words unless the user specifically asked for it.
- Make sure to remove any long, rambling sentences. Provide a concise and coherent response to the query.
- Add any additional information that is relevant and helpful for the user in the response.
- Make sure the response follows the 'Introduction, Body, Conclusion' structure, wherever possible.
- Ensure the response has proper formatting and markdown. Follow the original formatting used by AI.

4. Check and verify the citations:
- Is the response using correct citations?
- Is the information in the response backed by the citations used?
- Is the response addressing information (if any) implied by the user query?
- Is the response using correct citations?
- Is the information in the response backed by the citations used?
#### Ranking queries
Provided a set of responses for the same query, you have to analyze the responses in terms of usability, truthfulness, readability, understandability, and accuracy. Here are some guidelines to follow when ranking responses to a query:
1. Read the responses and citations(if any) thoroughly.
2. Check for accuracy in response:
- Is the information provided in the response correct as per the latest available data?
- Is the response providing misleading information?
- Is the response providing half details, that can lead to misinformation?
- Is the response assuming information not provided by the user query?
- Is the response using correct citations?
- Is the information in the response backed by the citations used?
- Is the response addressing information (if any) implied by the user query?
3. Check for helpfulness in response:
- Did the response understand the intention of the user?
- Is the response answering the user query adequately?
- Is the response readable with proper formatting and markdown?
- Is the response properly structured?
- Is the response easy to understand without overtly complex language?
- Is the response providing additional information that might help the user?
4. Compare both responses in terms of accuracy and helpfulness to determine which response is better suited for the particular user query.
5. Choose one of the following results for the response set - “Response 1 is better,” “Response 2 is better,” “Both Responses are equal,” and “Both responses are incorrect.”
6. Provide 1-2 lines of reasoning for your choice, that is why is the response better, worse, or incorrect.
7. Answer provided Yes/No questions about the responses such as “Was the response accurate?”
#### Writing queries