After spending so much time testing ChatGPT with every kind of request, I’ve learned where it tends to fall short.
It can hallucinate, some models rely on knowledge with cutoff dates that make recent events harder to handle, and it doesn’t always excel at complex reasoning unless the prompt is carefully structured. Because of that, I go into every interaction knowing I need to be specific and push the chatbot to give its strongest possible response.
