Quality Control Systems
👋 Hey there!
At Insight Goat AI, we prioritize the quality of conversations between your website visitors and our AI chatbot. To achieve this, we've carefully designed three quality control systems specifically aimed at enhancing the user experience.
To access these features, simply click on this link.
Here are the three systems that help improve conversation quality:
1. Prevent Hallucinations
Our first system addresses cases of low confidence. The AI evaluates the user's query and cross-references it with your Knowledge Base.
By doing this, the bot calculates a confidence score that indicates how well it can respond to the query. If the confidence score is low, instead of providing an uncertain answer, the bot will inform the user that it is unsure. To ensure your visitor receives assistance, it typically redirects them to your contact form.
2. Prevent Unrelated Conversations
The second system is crucial for maintaining the relevance of conversations. The AI checks whether the user's query aligns with the information available in your Knowledge Base.
If the user's question is unrelated to your business or Knowledge Base, the chatbot politely informs them that it cannot assist with that specific query.
3. Prevent Offensive Messages
Our third system is designed to ensure that conversations remain professional and respectful.
Using OpenAI's Content Moderation API, the chatbot screens user queries for rude or offensive content. If a query violates OpenAI's Terms of Service or contains inappropriate language, the chatbot responds with "I can't help with that" to maintain a respectful communication environment.
Side Effects
Please keep in mind that activating these Quality Control Systems will double the token usage per conversation. Additionally, due to the extra evaluations these systems perform, responses may take an extra second.
We hope this provides a clear understanding of our Quality Control Systems at Insight Goat AI. If you have any more questions, feel free to reach out!
Last updated