Higher score than computer? Are this batch of graduating artificial intelligence students really delicious?

Since the beginning of this year, artificial intelligence has officially become the cusp of the storm. The resources of large factories are running into the market, and emerging challengers are emerging one after another. Students who have just finished the college entrance examination are also surprised to find that many schools have new artificial intelligence majors, and the scores are even higher than traditional computer majors.

It is not uncommon for this craze to move from industry to universities. In 2018, 35 domestic universities took the lead in obtaining artificial intelligence professional construction qualification students. In the following three years, about 100 universities applied to offer this program every year, including undergraduate colleges and junior colleges.

For AI students graduating this year, it seems even more unusual. These young people who took advantage of the "dividend" were pushed to the forefront of the times in ignorance. How long will this wave last? What is the real employment situation? What are the career prospects? It was the season when college students graduated and volunteered for the college entrance examination, and they left with confusion and were replaced by another group of confused eyes.

One

Unknown means more possibilities

In July, as a teacher of artificial intelligence at a junior college in Shenzhen, Yan Hu was almost on his way to various admissions seminars. With the ChatGPT craze, many students are curious about this field, and he receives a large number of calls from parents every day.

According to incomplete statistics from Nandu reporters, there are currently 502 undergraduate schools and 515 junior colleges in China that have opened artificial intelligence. This year alone, a total of 116 universities applied to offer artificial intelligence majors.

Yan Hu introduced that the training direction of undergraduate and junior college majors in artificial intelligence will be different. Specialist courses focus on the operation and maintenance deployment of models, while undergraduate courses should delve into the algorithm principles behind them, giving students more possibilities for further study and research and development.

Behind the blossoming everywhere is the market's optimism about the prospects of the industry. Zhaopin's recruitment report pointed out that artificial intelligence has become a high-potential industry with broad prospects relying on high technical content and the support of national industrial policies. Especially with the advancement of science and technology, the development of intelligent manufacturing has brought new business growth points to enterprises, and enterprises are eager for high-end technology and R&D talents.

This year's popularity of generative artificial intelligence has also triggered a wave of "recruitment" of Internet manufacturers. On recruitment software, artificial intelligence high-paying positions released by companies abound, which are particularly bright in the bleak job market.

This naturally triggered a wave of high scores for filling in artificial intelligence majors. In addition to students, there are also teachers - Yan Hu recently completed a teacher recruitment fair, and artificial intelligence doctors from many universities came to inquire.

For parents, the unknown means the greatest fear and the greatest hope. As an emerging discipline, its employment direction, market demand, and the combination of courses and practice are still in the fog. Many parents can only ask everywhere with trepidation, is it really easy to find a job in a new major? Can this year's boom continue four years from now?

Two

Do you need to study for a PhD to get a job?

Taking Yan Hu's school as an example, Liu Fu is the first batch of college graduates majoring in artificial intelligence. After graduation, she entered a small and medium-sized technology enterprise with AI+ intelligent manufacturing in Shenzhen as a machine vision engineer. This is an industry that has only emerged in recent years, but it has a wide range of applications.

Liu Fu recalled that when he first filled out the volunteer, he didn't have a deep understanding of artificial intelligence. Because I was interested in robots since I was a child, I thought that the artificial intelligence major was to learn how to manipulate robots, and I felt that it was particularly amazing, so I chose this major.

Later, it turned out that artificial intelligence was much more than that, and three years of study was not enough for her to have a deep understanding of robots. Her major is branched out from the traditional software profession and is divided into three directions: computer vision, using computers instead of human eyes to do object detection and other work; Natural language processing, including machine translation and text generation; Speech processing technology to assist machines and humans to interact with sound.

Compared with the computer major that focuses on programming ability, Liu Fu's professional course emphasizes the cultivation of algorithm model deployment ability, including some data processing courses. The school attaches great importance to the combination of theoretical knowledge and practice, and will arrange some practical training projects. Like the very popular ones in recent years, she has been exposed to related projects such as face recognition, mask detection, license plate recognition, and natural language processing.

At present, her position is in line with her professional direction, and her daily work is to build a vision implementation plan and use machine vision to accomplish things that the human eye cannot do. For example, the robotic arm on the assembly line can be quickly positioned and directed by machine vision to label products, high-precision measurement of objects with an accuracy of 0.1 mm and below, and defect detection of metal surfaces in the intelligent industrial field.

Of the nearly 80 students who graduated with Liu Fu, 1/3 chose to go on to higher education, 2/3 were directly employed, and most of them could find jobs. There are two main lines of employment, either engaged in traditional software development, software testing and other work, or machine vision, model training, data labeling and other positions strongly related to artificial intelligence. Overall, Liu Fu feels that there are still relatively few students engaged in R&D work. Positions such as software development or algorithm research and development tend to have higher thresholds and prefer computer majors.

To say that it is not difficult to find a job is fake. Liu Fu had delivered to many companies at that time, not limited to artificial intelligence, but also found software testing, software development, and even some clerical jobs. The first thing to face is the problem of academic qualifications, there are many and wide range of things that need to be mastered in the field of artificial intelligence, and junior college students will still face some restrictions in terms of academic qualifications. The second is gender, especially in the field of AI+ intelligent manufacturing, employers are more willing to choose men. Including her current job, from time to time customers and peers will say that it is rare to see female machine vision engineers. Finally, as a fresh graduate, I have no experience, the things taught in school are more basic, and the practical content is insufficient.

Yan Hu also admitted that the corresponding positions of artificial intelligence majors overlap with traditional software development and data analysis. In particular, a large number of startups have emerged in the field of artificial intelligence, which are more inclined to recruit high-end talent in the short term.

This is also the question that many parents of college entrance examination students consult him, does it have to study for a master's or doctoral degree to find a job? As things develop to a certain stage and the division of labor is further refined, Yan Hu believes that more positions suitable for talents at different levels will inevitably appear in the field of artificial intelligence. When the industry matures, undergraduate graduates will also have good competitiveness.

Yan Hu believes that the greatest potential of artificial intelligence lies in its application scenarios. Not long ago, it was vision technologies such as face recognition, and the rise of generative models this year caught everyone by surprise. Diversified landing places give people more imagination, and will inevitably bring more recruitment needs.

Three

"The industry doesn't need that many people yet"

The strong recruitment demand needs to be supported by broad market prospects. Xiao Wang, who just graduated from a top 985 university in Beijing this year, is not so optimistic about the competitiveness of artificial intelligence undergraduates or junior college students in the job market at this stage.

After graduating with a PhD in artificial intelligence, Xiao Wang will join a start-up to work in the research and development of artificial intelligence algorithms. This year, there are indeed many companies looking for artificial intelligence students, and he has also received many offers from large manufacturers, but it is obvious that enterprises have high requirements for artificial intelligence algorithm engineers, and are in a state of rather shortage than abuse. Even if there is a rush to "recruit", what is urgently needed is high-end talent.

In the process of finding a job, he can see that some general practitioners in the field of artificial intelligence are struggling in employment and are difficult to find satisfactory jobs. The person in charge of a large factory asked him to help recommend candidates, saying that although AI is popular this year, it is still difficult to recruit people, and the technical level of candidates is relatively high.

The industry is highly competitive, one reason is that the algorithms of artificial intelligence are relatively fixed, and it really does not need so many people. One trend of large models is to make the model bigger and bigger, using more and more data. When the computing power is sufficient, the algorithm of the model does not need to be adjusted. Because studies have shown that the adjustment of algorithm and model structure has little impact on the overall performance. Therefore, even large factories will not invest too much in the exploration of algorithms.

What are so many people going to do? You may work on data, become a data labeler or data cleaner. The reality is that these jobs do not require strong professional skills, and low technical barriers lead to higher substitutability and greater competitive pressure.

Xiao Wang bluntly said that even if there is a surplus of talents in the industry in the future, he will not reject it if he is needed to do work such as data cleaning. Because as a practitioner, he knows how important data cleansing is. But for the original practitioners, they have to face competitors with higher backgrounds.

"My understanding is that there are not so many middle-level subdivisions in the industry at the moment. Bottom-level jobs do not require too high years of education, and competition for more top-level R&D positions is also very fierce, and graduates of ordinary colleges and universities cannot grab it. In his opinion, the artificial intelligence industry does not need so many people at present. A lot of teams do great things with a very small number of very good people and a lot of money.

Of course, this is only part of the speculation. Like everyone, Xiao Wang cannot easily conclude the future of artificial intelligence. If it becomes a skill that everyone needs to master like computers, then the market demand for talents will be very broad, and the division of labor will become more and more refined. For example, small and medium-sized companies will need a large number of artificial intelligence practitioners to help them develop their own products on the basis of the models trained by large manufacturers, which will be a very wide range of practice scenarios.

Seeing that many schools have opened this major this year, and the professional recruitment score is still very high, Xiao Wang feels that students still need to understand and judge in many aspects before making a choice, and avoid simply pursuing dividends in four years based on the current dividends. When submitting resumes, it is ideal for interviewers to treat artificial intelligence and computer science students on an equal footing. But if they think that undergraduate artificial intelligence is too subdivided and programming strength is not as good as computer students, it is difficult to say how artificial intelligence students will deal with themselves.

Really interested in this direction, Xiao Wang feels that the best way to learn is to master the basic knowledge through books, and then go to the most cutting-edge papers, or follow a capable teacher to do scientific research. Schools where teachers are not so good, and internships can be done through various channels. In contrast, he believes that classes are not so important, especially since the field of artificial intelligence is now making new discoveries every day, and each new discovery is changing the way practitioners think about the industry. If the course content does not keep up with reality, the knowledge acquired in the classroom is quite limited.

Xiao Wang is a computer science student, and after a long time of exploration in the laboratory, he decided on artificial intelligence as a research direction. However, he stressed that if the time is not very generous, it is not recommended to enter the laboratory at the undergraduate level. Because artificial intelligence is a cutting-edge exploration of research, it is best to have a teacher one-to-one or one-to-many zone. If you don't bring it, you may spend a lot of time but you can't do anything, or even enter the real scientific research state. Especially in some schools with weak teachers, the energy and ability of teachers to lead students are limited.

It is more important to lay a good foundation in computer science or mathematics at the undergraduate level. Subdivisions such as artificial intelligence can be put into graduate school. Because artificial intelligence covers more and more fields, it can be studied from different directions such as language, biology, and medicine, and it is difficult for undergraduates to take into account so much.

"This does not mean that the content of artificial intelligence is not broad enough or deep enough, but it has not yet formed a complete theoretical system." Regarding artificial intelligence, there are still many opinions in the academic circles, and there are many academic views. Only through experimentation and research can students discern the difference. Many undergraduates may not have enough energy and no conditions to do experiments.

Four

Internship challenges for undergraduates

Not only this year, in fact, in recent years, artificial intelligence has been unabated. In 2021, the 2021 college entrance examination of Xiaoshan, the artificial intelligence recruitment score is even higher than that of computers. At that time, this Sino-foreign joint venture school in the south opened an artificial intelligence undergraduate major for the first time, and the parents thought that the school would be more attentive to the first batch of students, and this topic was particularly hot, so they chose it.

Oyama High School had a little programming and wanted to study computer related majors in college. After being admitted, I consciously searched for artificial intelligence news and understood the meaning of artificial intelligence.

After going to university, the school's attention was not felt, and Xiaoshan felt more obvious that many professional courses and computer majors were the same, and teachers were also shared. The basic courses such as mathematics and algorithms she studied in her freshman and sophomore years are not highly related to artificial intelligence, and the instructors are not necessarily artificial intelligence majors, which is equivalent to laying the foundation for subsequent learning.

Open the professional introduction of the school's public account, most of the teachers are computer majors, and they are in line with artificial intelligence in their research direction. It is not difficult to understand that artificial intelligence is an emerging discipline, and the time for colleges and universities to establish departments is often not long, and it is conceivable that there are not enough talents to fill positions in the short term.

It's worth noting that this is not a school's dilemma. Yan Hu has communicated with many teachers of the same major in brother colleges, and generally reflects that the biggest problem is the problem of teachers, it is difficult to have teachers who understand artificial intelligence and teaching to support this major, and there are not enough doctors. Among the only doctorates, very few are willing to come to universities, after all, companies offer much more generous salaries.

"If you want to go in the direction of algorithm research, the competition will be very fierce. In our school, we may have to work our own. Although there was still some time before graduation, Oyama already felt the anxiety of going to graduate school. Not only is it difficult to find a job after graduating from a bachelor's degree, but even internships are required to publish papers and scientific research results.

Judging from Xiaoshan's experience of looking for an internship, he was basically eliminated at the resume level. Many interns of artificial intelligence algorithms require a bachelor's degree or above, and their published papers are preferred. And judging from the written test questions, even if you finally choose, it is quite difficult to complete the corresponding work. Compared with the internship positions corresponding to computer majors, artificial intelligence positions are generally more demanding.

Although he has been taking professional classes for two years, Oyama feels that there are very few experiences that can be put on his resume. Most of the internships shared in the professional group are public internships, such as campus ambassadors, etc., which are not related to artificial intelligence. The school's teacher assistants are happy to provide students with research opportunities, and several laboratories welcome undergraduates. This summer, Oyama also hurried to join a teacher's project to study the use of artificial intelligence algorithms for DNA sequencing.

The Nandu reporter learned that some schools are trying to solve the problem of insufficient internship resources in the form of school-enterprise cooperation. Including joint creation of holiday training courses with local enterprises, with products as the carrier, the practical application scenarios of technology are transferred to the school classroom, and solutions are proposed by students.

But in any case, in Oyama's option, undergraduate direct employment is not considered from beginning to end. So for enterprises, is there a hard threshold for academic qualifications? The head of recruitment of a leading artificial intelligence company told Nandu reporters that different positions in the labor field have different requirements for the professional background and basic ability of practitioners. For example, algorithm researchers have higher requirements for practitioners, and generally speaking, most of the algorithm researchers we see are doctors or masters.

The main positions of artificial intelligence students are computer vision, automatic driving, large models, R&D engineers, etc. The above-mentioned person in charge said that the interviewer valued the most, one is engineering ability, familiar with infrastructure and engineering skills, able to carry out architecture design, and implement scientific research problems. The other is scientific research ability, learning to acquire domain knowledge, research to define scientific research problems, and propose innovative solutions. Therefore, it has strong programming ability, learning ability, and problem-solving ability, and is more favored.

He suggested that as a student majoring in artificial intelligence, you can actively participate in some scientific research projects during your school and communicate with scholars in various research directions. If conditions permit, intern in major enterprises, combine theory and practice, understand your interests and enterprise needs, and make career planning earlier.

(At the request of interviewees, "Yan Hu", "Liu Fu", "Xiaowang" and "Xiaoshan" in the text are pseudonyms.) ）

Written by: Nandu reporter Huang Huishi, intern Zhang Yuxuan, Zhang Yaju, Nandu reporter Hu Gengshuo