How to better adapt to ChatGPT?

28 Haz 01:20

When I first tried ChatGPT, I immediately thought of Google Assistant or Siri’s basic command understanding—but the difference became clear very quickly. When I asked Siri, "How’s the weather?" I’d only get a word-based answer, whereas when I gave ChatGPT a detailed prompt like, "What’s the weather in Istanbul today, what are the best hours for photography, and what’s the nearest location I can reach with a maximum of 3 kg of equipment?" it understood the full context and provided an optimized response. So instead of simple commands, you need to "build a narrative"—almost like explaining a project from start to finish to an assistant. At first, my prompts were as simple as "Write me an article," but the results were either too superficial or unnecessarily long. Then I realized that just like I do with Midjourney when generating images, I should set specific criteria for ChatGPT too—like, "Write a 150-word blog post with three main actions, avoiding financial jargon, targeting young entrepreneurs." The outputs became much more focused. Also, don’t forget that unlike Siri, its biggest limitation isn’t just the lack of real-time data access—it’s the "default knowledge cutoff (end of 2023)." So don’t be disappointed if you ask for yesterday’s stock market data; it’d be like asking a newspaper for traffic conditions from 15 minutes ago—just as absurd.