How Knowledge Graphs Reveal Bias in LLM: ChatGPT & Claude Extension

In this video, I'm going to show how to use https://infranodus.com to detect and reduce bias in ChatGPT & Claude conversations using knowledge graphs and text network analysis. The approach is based on identifying the main topical clusters in text and revealing the topics within. Then we estimate if some topics or concepts have too much influence on the discourse and focus on the less represented clusters of ideas to develop them further using the structural gaps between them. You can use it on your existing ChatGPT conversations with the InfraNodus browser extension — https://infranodus.com/extension — or in Claude using the InfraNodus MCP server — https://infranodus.com/mcp Timecodes: 0:00 Using graphs to detect bias 1:52 Analyzing ChatGPT conversation 2:44 Identifying excessively influential topics 3:57 Reducing bias in text using the content gaps 5:27 Diversifying your LLM’s thinking with gap questions 5:54 Using MCP to reduce model’s bias 7:52 Installation instructions Try it for free by installing the extension or MCP server directly — now without even opening an account! For advanced workflows, create a trial account at https://infranodus.com