To deal with AI, you need to personally deploy and direct it It has to be said that this AMM challenge is indeed interesting, even a bit addictive. At present, the score is only 450+, far from the top 100 score of 520+. But the process of exploration is quite interesting. The current AI models are very good at coding, fine-tuning, and parameter optimization, and can be called self PUA masters. For example, from a score of 380+at the beginning, progress to 400+. This is what it excels at. But it is also easy to fall into a vicious circle and not be able to break out. Someone tweeted that it's easy to score 450+, but in reality, I find it quite difficult (perhaps because I'm just too inexperienced) ）。 Human users need to help AI: broaden their thinking. If not, if you expect AI to iterate crazily and make the score climb to 500+, I think the possibility is extremely low, even if you stack Opus 4.6, CodeX 5.3, Gemini 3.0. You need to intervene in a timely manner and give AI a great direction. This requires you to have your own directional thinking. Going up, if you really want to make it to the top of the list, you need to have a certain understanding of computer science and software engineering in order to achieve a golden combination: Human users pointing in the right direction+AI frantically trying, local optimization=ultimately achieving the 'optimal solution' Even if the level of AI intelligence continues to evolve in the future, it will still be the most programming savvy individuals who can extract maximum value from AI. The top 1% of people have achieved productivity improvements of 100 times or 1000 times; We ordinary people have gained a tenfold increase in self-efficacy. Others can enjoy the digital experience of "nipple joy" produced by AI with peace of mind. Everyone has a bright future.