最近DeepSeek的出现，又把AI应用推向了一个新的高潮。我一直在想这个问题，通过AI工具如Chat GPT搜索出来的结果，有多少可信度呢？我们如果仅仅是好奇而查阅一下某个问题，并不把结果当真，倒也无所谓。但如果是想要把结果用来支持某个研究，就要认真想一想了。最好的处理办法是顺着线索把有关问题再查证一遍。即把AI工具当作一个助手，而不是完全依赖它。
我现在做了一个实验，就查
Review of XPress-MP，分别用Chat GPT和DeepSeek查，发现了结果很不一样。两者的口气不一样。
DeepSeek: "Okay, so I need to review XPress-MP. Let me start by recalling what I know about it. XPress-MP is an optimization software, right? It's used for mathematical programming, like linear programming, integer programming, maybe even nonlinear problems. I think it's developed by FICO, which used to be called Fair Isaac Corporation.


First, I should outline the key features. It's a solver, so it has algorithms for different types of optimization problems. Maybe it's known for handling large-scale problems efficiently. Users might be in industries like logistics, finance, manufacturing. Let me check if there are specific industries where XPress-MP is popular.


I remember that XPress-MP has a modeling language, maybe similar to AMPL or GAMS. That allows users to formulate their models in a more readable way instead of coding everything from scratch. The Mosel language comes to mind. Mosel is part of XPress-MP, right? So users can write models using Mosel, which then interfaces with the solver. That's a plus for usability."


结果也不尽一样。


结论就是，最好只是把AI工具当作一个助手，还是需要自己动手才放心。