I'd almost lost interest in AI when I stumbled upon Fu's tweet. Using a prompt, I generated a web-based version of Honor of Kings using Claude's Fable 5.2. While it has some flaws, Fable 5.2 is definitely better than I expected. However, after careful analysis, I realized I'd fallen into a trap: this game isn't original; it's essentially a derivative work. PS: I wonder what level of effort Fu put in. Following his instructions, he worked on it all night, resetting it twice—equivalent to a three-week budget, likely high or even extremely high. This is essentially burning money. Although it looks like one night, it's actually using money to cram three weeks' worth of work into one night. There's plenty of open-source code similar to Honor of Kings online; the structure is already quite clear. The AI's job is simply to copy it, which makes it seem very powerful. But for a completely original game, it would be exhausting. But this is a good direction: build a foundation first and then gradually improve. Let's try again.