Abstract:
In the fall semester of 2026, a course in the School of Computer Science at Nanjing University exploded onto the Internet with a "hard-core opening statement."
Jiang Yanyan, associate professor and doctoral supervisor of the college, made a striking point in the first lesson of the new course "Generative Software Engineering" - "CS students without tokens should drop out of school immediately." This sentence was highlighted in red and placed on the homepage of the courseware. It was extremely impactful and swept the screen in technology circles and social platforms.
It is understood that Jiang Yanyan has a solid reputation and high professional recognition in the domestic computer science community. The "Generative Software Engineering" offered by it is a brand new course in the fall of 2026. The biggest highlight is the introduction of the Token self-funded training mechanism, which completely subverts the free tool usage model of traditional computer classrooms.
According to the course rules, all practical training and assignments this semester will be completed based on the deepseek-v4-flash large model. Students need to pay for tokens to support experimental and development tasks.
Jiang Yanyan explained in class that the core purpose of this move is to break the "zero-cost thinking" formed by students' long-term use of free AI tools and truly establish the three core awareness of cost, quality and delay that are crucial in engineering development.
In the teaching system of this new course, Token is officially defined as experimental consumables for computer majors in the AI era. It is equivalent to reagents for chemistry majors and experimental equipment for physics majors. It is an indispensable basic resource for modern software engineering practice.
At the same time, the course clearly prohibits the phenomenon of "tokenmaxxing", strictly eliminates meaningless and repetitive waste of computing power, and requires students to accurately and efficiently use large model capabilities to solve truly valuable engineering problems with limited tokens.
Regarding the controversial statement of "drop out without a token", the outside world generally believes that this is a typical exaggerated teaching warning and is not a true punishment for dropping out.
People who are familiar with Jiang Yanyan’s teaching style said that his teachings have always been outspoken and hit the pain points. The essence of his remarks is a sharp reminder: AI has fully penetrated the entire process of software engineering. If computer students still stick to the old model of traditional hand-written code, answer questions and take exams, and refuse to embrace large model tools, they will completely fail to keep up with the speed of industry iteration, and they will not be able to close the gap with the forefront of the industry just by staying up late and working hard.
It is worth mentioning that the key to the smooth implementation of this self-funded training model lies in the cost-effectiveness of the model selected.
It is reported that the price of deepseek-v4-flash is only a few tenths of that of the GPT series. The cost for individual students to purchase Tokens is extremely low, which will not cause financial pressure. It also makes "self-funded training" realistic and feasible, and avoids the problem of disguised dissuasion caused by high-priced computing power.
In fact, behind this Internet craze, it also reflects the transformation trend of computer education in domestic universities: the ability to use large models and implement AI projects has been upgraded from bonus skills to core competencies for computer major students. What do you think about this?
Comments