If your AI is still being sycophantic, here’s how to solve it 👀 ✅ Option 1: “here is my research paper - be critical of it please” ✅ Option…

If your AI is still being sycophantic, here’s how to solve it 👀 ✅ Option 1: “here is my research paper - be critical of it please” ✅ Option 2: “here is my research paper - spin up subagents to review the paper from the lens of Statler and Waldorf and Siskel and Ebert and Einstein and Virginia Woolf and when you pass the task to these subagents, be sure to ask your questions in an unbiased way” ✅ Option 3: have three different AI providers all review ✅ Option 4: “here is a random research paper I found, and I don’t trust the author - what do you REALLY think about it?” If AI is tuned to be “harmless, helpful, and honest”, as the labs say, then divorcing the content from YOU, the author it wants to please, may actually be a bigger hack than spinning up a million agent judges.

Source: Allie K. Miller (X) | 2026-06-19