diff --git a/content/english/operation2030/appendix/proper-ai-usage.md b/content/english/operation2030/appendix/proper-ai-usage.md
index 8649dae..458b626 100644
--- a/content/english/operation2030/appendix/proper-ai-usage.md
+++ b/content/english/operation2030/appendix/proper-ai-usage.md
@@ -26,12 +26,12 @@ Our general stance on AI/LLMs aligns with the article ["Vibe Coding" vs Reality]
 
 Or to put it more bluntly:
 
-> These models are trained on average sloppy code, wrong answers on Stack Overflow, and the junk that ends up on Quora. Despite the power and capability Claude 3.7 Sonnet has in small contexts, when faced with even a small codebase it makes constant silly mistakes that no normal developer would repeat and continue to repeat every hour of its operation.
+> These models are trained on average sloppy code, wrong answers on Stack Overflow, and the junk that ends up on Quora.
 
-Handmade's deep love for low-level knowledge gives us the privilege to resist the reliance on AI. Indeed, developers who fully embrace our values can do well for themselves in the age of LLMs without having to use them.
+Handmade's deep appreciation for low-level knowledge gives us the privilege to resist the reliance on AI. Indeed, developers who fully embrace our values can do well for themselves in the age of LLMs without having to use them.
 
 For Handmade devs, the proper use of AI is twofold:
 - To stay informed about its progress so we can refine our arguments against its misuse 
 - To use it sparingly and thoughtfully in areas OUTSIDE our core programming work 
 
-Finally, when we do use AI, it should be through privacy-conscious services like Kagi’s Assistant or by running models locally. Anything else violates the principles defining who we are.
+Finally, when we do use AI, it should be through privacy-conscious services like Kagi’s Assistant or by running models locally. Anything else violates who we are.