I have a hard time trusting the responses I get from AI models. Recent releases have become far more accurate, and hallucinations have reduced considerably. That is a welcome change, but I still don’t trust a standalone AI response as much as something I can verify through Perplexity.