After DeepSeek became popular, the discussion on "AI makes liberal arts useless" was once again ignited. The topics related to "DeepSeek Liberal Arts" have been on the hot search several times.

Many people believe that big models like DeepSeek can already create high-quality humanistic works and can also obtain massive humanistic knowledge online. AI is already capable of writing official documents, writing papers, and creating copywriting, and there are other big models of Wensheng pictures, Wensheng videos, and Wensheng music, which can replace art and design work.

Looking at this point, it seems that it is easy to conclude that AI is about to replace liberal arts work. It seems that choosing to study humanities majors, or even choosing liberal arts and sciences majors is a self-destructive path. The sense of panic brought by this public opinion pressure has spread to the group of liberal arts students and liberal arts parents, covering this spring of AI with a layer of yellow sand that seems to be burying humanistic education.

But, is that really the case?

Every time AI technology is popular, someone will repeatedly talk about the "useless theory of liberal arts".

But this round of AI climax brought by DeepSeek just made me see the importance of humanistic abilities in the AI ​​competition, and even the employment prospects of liberal arts students in the AI ​​era.

You might as well think of a question: It is said that AI makes liberal arts useless, so does AI big model need liberal arts students?

Give some examples, you will find that humanistic ability is crucial in today's big model competition, and it even determines the direction of big model competition.

First of all, let’s talk about DeepSeek. If you compare a few big models a little, you will find that DeepSeek has little gap with other big models in basic capabilities such as information collection, and even the problem of big models hallucinations is a bit serious. But why can it continue to be popular around the world? In addition to industrial factors such as reducing AI hardware costs, the most important thing is that it has excellent topic dissemination.

Compared with other large models, DeepSeek's answers have a better sense of network, a stronger sense of humor, and a more beautiful and poetic writing style. These abilities lead to a stronger "real-person feeling" answered by DeepSeek under the same question. This leads to users who are more willing to talk to them in terms of interest and are more willing to share the results of the conversation on social media.

So, do these humor, sharpness and poetry come from some kind of algorithm optimization? I'm afraid not. At least the careful selection of model training corpus will play a key role in it, and this grasp of literary, philosophical, humorous, and even social communication is probably the home court for liberal arts students.

There is an example that can particularly strongly demonstrate that DeepSeek attaches great importance to humanistic abilities and has definitely supported liberal arts students behind it. Ask DeepSeek some macro questions, it will say some words that are particularly grand and far-reaching but have no practical meaning. Such as "genetic cube" and "phantom cage" are very similar to humanities majors that cannot be written.When I was a essay, I made up a random routine, and I was very frightening and useless, and I would never be severely investigated.

When looking back at DeepSeek's rapid progress, you will find that it is the emphasis on humanistic abilities that make it stand out on the public side. The so-called "a war of hundreds of models is the same." One of the reasons for the serious homogeneity of big models in the past was that it focused on algorithms and neglected content, and ignored AIGC in essence, content creation. Recently, the AI ​​industry has been discussing reducing the cost of training like DeepSeek, but if we do not see its emphasis on humanistic abilities and content generation, most subsequent big models will still not be able to escape the boring content and the boring dialogue, and will be immediately abandoned by users.

A few other examples can be added to discuss the importance of humanistic abilities to large models. I found that friends who work as civil servants and teachers like to write official documents with just one word, especially work summary, meeting minutes, etc. Their reason is that Wen Xinyiyan is more appropriate and has a neat format than other big model terms. This is very important in public jobs and can also reduce the amount of subsequent revisions. From this perspective, the understanding of liberal arts work such as document writing has also become the key to anchoring the user population and improving one's own competitiveness.

Another example can be taken at the Wensheng Picture Big Model. Until today, we all lament that Midjourney has no substitute in the country. Even with difficult operations, complex prompt words, and lagging networks, users with relatively professional needs are still flocking to Midjourney one after another. The fundamental reason is that the aesthetics of a large number of domestic literary and artistic graphic models are too poor, and the generated pictures always reveal an indescribable earthy smell at the end. The reason behind this is the lack of high-quality paintings and design works for AI to learn. The liberal arts abilities such as painting, design and aesthetics have seriously limited the development of literary pictures, literary videos, and even the entire multimodal route.

Who says that liberal arts cannot determine AI productivity?

To some extent, AI will not make liberal arts useless. Without liberal arts students, AI may be useless first.

In fact, the argument that "when a certain advanced technology comes, the liberal arts will be useless" has long been a cliché. I remember when I first appeared on the Internet, some people said that you can find liberal arts knowledge in the future. Is rote memorization useful in learning liberal arts? Later, the media came, and some people said that anyone could write in self-media. Is there any way out for those liberal arts students who work in newspapers and magazines?

The story later was exactly the opposite. The Internet industry and self-media have become a strong growth engine for liberal arts employment. Even some liberal arts majors that once had a very narrow employment range, such as archival and museum, can use popular science to use self-media as a new career opportunity. It is said that when short videos come, no one can read the text. However, short drama screenwriter, video planning and other work have also absorbed a large number of liberal arts jobs. With the advancement of technology, liberal arts students not only have a wider employment range, but are relatively more likely to enter a high income.

Today, the big model seems to be coming to squeeze againI lost my job as a liberal arts student. But it is easy for us to ignore that the content of the big model does not come out of thin air, but also requires humanities to provide training corpus for it and supervise it for effective content generation. On the other hand, big models can also become new content creation assistants and workplace support like self-media platforms.

Even in the early stages of the outbreak of the big model, a large number of new liberal arts employment directions are coming, such as:

1. The AIGC content officer behind the big model.

DeepSeek's success has shown that content generation quality will be a key element in determining the success or failure of a big model. Next, we will surely see the new trend of AI manufacturers paying more attention to humanistic abilities and high-quality corpus screening. In this process, the training data screening work that could have been outsourced at low cost will become a high net worth work led by practitioners of humanistic backgrounds. AI content officer is probably not far from us.

2.Agency developer.

What is the biggest obstacle for liberal arts students to become software application developers? Probably not able to program. Well, now AI can do it. While the big model eliminates the threshold for content generation, it is also eliminating the threshold for technical positions such as code. Application development driven by humanistic background is likely to become a new explosion point for employment and entrepreneurship.

3. Multimodal content creator.

By the same, it is difficult for a liberal arts student to turn knowledge, theory, and literary thoughts into images. Because it requires technical capabilities such as cross-professional to the art field, and even video shooting and editing, these AIs will also do it. After literature, pictures and videos, the next liberal arts employment trend may be multimodal creators.

Summarize these new opportunities and we may reach a consensus: AI brings not discipline discrimination, but fair opportunities.

Whether it is filling in bills, factory quality inspection, or writing code, copywriting generation, or writing report summary. From today to the future, AI can only replace simple and repetitive tasks. So the jobs it can replace are also concentrated in this field, namely those simple, highly repetitive jobs and positions.

At the same time, AI can do simple work, which also means it can make up for everyone's shortcomings in ability. Pure code workers can rely on big models to obtain systematic humanities and social science knowledge; people who have never been exposed to artistic creation can use AIGC to improve their aesthetic and design capabilities; an entrepreneur with a complete liberal arts background can also rely on agents and codeless development tools to become an AI developer, stepping into the trend of the times.

The biggest impact of the AI ​​era on individuals is not that a certain discipline is useful or a certain discipline is useless, but that we require us to reorganize our own ability system: everyone must have creative abilities that cannot be replaced by AI, and at the same time learn to hand over those jobs that they are not good at and lack interest to automation and intelligence, so as to achieve more comprehensive competitiveness and more fair social opportunities.

Technical ToolsThe upgrade has always only made the world open and inclusive.

From this perspective, intelligence is fairness.