My First Blog

This blog is my first blog. I tried to document some random (superficial but might be intersting) thoughts about AI, which were orignally in Chinese but translated to English by Gemini 2.5 Pro, as you will see below.

Translation by Google Gemini:

After delving into cutting-edge statistics papers on topics like Causal ML and Conformal Inference during my junior year, I’ve found myself wondering: why are statistics enthusiasts like me, and even established academics, so increasingly drawn to the field of AI?

Given the limited space and my own limited understanding, I’ll offer just two perspectives to hopefully kickstart the discussion.

First, from the perspective of its societal utility. It goes without saying that AI is delivering an unprecedented impact on human-centric sectors like medicine and education, on engineering disciplines that serve our material world (such as CS and ECE), and even on creative domains founded on the human intellect—like art, math, philosophy, and literature—which are built on rational deduction and emotional creation. It is fundamentally reshaping how we live and work, while simultaneously creating a host of new scientific and philosophical questions. These new frontiers are, without exception, captivating the curiosity of both amateurs and experts alike.

Second, on the linguistic front. The global phenomenon of AI tools seems partially linked to a kind of “textual aphasia” (文字失语). People are finding that AI’s expressions can be more vivid, nuanced, and witty than their own, which indirectly suggests that modern human expression may be regressing or becoming homogenized.

However, we must first recognize that LLMs like DeepSeek, ChatGPT, and Gemini are built upon vast, diverse, and high-quality corpora. Their very construction represents a crystallization of collective human wisdom, operating on a scale that far surpasses the reading volume of a single human lifetime. For the vast majority of people, their linguistic output could hardly compete with this, “textual aphasia” or not.

And while people can often still detect a distinct “AI flavor” in the output, this high-quality, albeit “AI-flavored,” content can perhaps offer new avenues for creative work. We shouldn’t be so quick to dismiss it as mere “smoke and mirrors” (装神弄鬼的话术).

Original Chinese:

在大三读了一些统计前沿论文（例如Causal ML、Conformal）之后，我不禁好奇为什么包括我在内的统计学科研爱好者，甚至是高校里的统计学家，都逐渐被AI所吸引，由于篇幅有限以及本人认知的局限性，我只从两个角度抛砖引玉。从社会工具属性上，不必多说，AI对于医疗、教育此类以人为本的行业以及工程（例如CS和ECE）此类服务于人类物质世界的专业，甚至是对于艺术、数学、哲学、文学等以人的精神世界（例如理性推导与感性创作）为基础的创造性领域，都有着人类历史上前所未有的冲击——改变着生产生活方式，同时也创造了诸多科学和哲学问题，这些问题无不使专业领域的爱好者和专家感到新奇。从文学属性上，诸多AI工具风靡全球实际上和文字失语有一定关联——人们觉得AI的表达比自己的更生动更细腻更风趣，侧面反映出现代人的表达是退化、同质化的。但是首先DeepSeek、ChatGPT、Gemini等LLM都有庞大、多样、优质且海量的语料库和信息源支撑，其搭建本身就集结了人类的智慧结晶，而其规模又远超人一生的阅读量。对大多数人，即便没有“文字失语”，他们的语言表现也难以比肩。不论内容本身质量高低，人们仍能分辨出AI的色彩，不过这些高质量但“AI味”十足的内容或许可以为语言表达等创造性工作提供新思路，倒不必将其贬斥为装神弄鬼的话术。