166: Chapter 166 Loss of Faith

In the video, the foreign engineer's demonstration reached a shocking peak.

He casually said to the microphone: "Build me an e-commerce website with personalized recommendations, online payment, and user review functions."

On the screen, code poured down like a waterfall, but it wasn't just code.

Architecture diagrams, database schemas, API documentation, frontend component trees, deployment scripts.

"The entire process, from nothing to a running prototype, in less than half an hour."

A long silence.

Zhang Qian's fingers unconsciously dug into her palms, her voice trembling.

"This... this isn't writing code anymore, this is describing software. What value is left in our... our years of experience?"

At another table not far away, the conversation between two college students was also filled with confusion.

"Senior, I... I'm a bit scared."

The bespectacled boy, a senior AI major about to graduate, had a bitter tone.

"I've spent four years gnawing on machine learning theory and parameter tuning techniques, thinking I was a scarce talent. But now... a single model from them can automatically complete feature engineering, model selection, and hyperparameter optimization, and the results are far better than what we tune manually..."

His classmate tried to comfort him.

"Don't think too much, someone still needs to understand and maintain these systems, right?"

"Maintain?"

The bespectacled boy shook his head violently and pulled out his phone to open a news article.

"Look! Their latest system already has the capability for automated log analysis, root cause location, and even online hot-patching! It discovers its own bugs, patches itself, and continues seamlessly! What are we going to maintain in the future? Handing a wrench to an AI?"

On the same day, in Kyoto, a conference room at a top university's computer science school.

The atmosphere was so heavy it could be wrung out like water. An emergency meeting about "AI Impact and Teaching Responses" was underway.

"The students' anxiety can no longer be suppressed."

A young associate professor named Li spoke with urgency.

"Yesterday in my 'Data Structures' class, a student raised their hand and asked me directly: 'Teacher, GPT-5 can instantly generate optimal solutions for these complex balanced binary trees and graph algorithms, why do we still have to spend a whole semester bitterly trying to understand and hand-write them?' I... I didn't know how to answer to convince him."

The dean furrowed his brows and tapped his fingers on the table.

"Nonsense! Basic knowledge is the foundation! It's what lets you understand the essence of computer science and cultivate computational thinking! How can you deny the learning process itself just because the tool is powerful?"

"Dean, we understand the reasoning."

Another professor, Professor Wang, who is in charge of 'Software Engineering,' interjected, his face full of helplessness.

"But the reality is that AI is reshaping the definition of software engineering. The traditional waterfall flow of requirements analysis, design, coding, and testing is being overturned. There is already a huge gap between what we teach and what students will face in the future. We ourselves are already heavily dependent on these tools in our scientific research, so how can we righteously ask students to return to the primitive ways?"

An elderly professor with graying temples, Professor Liu, sighed and said something that silenced the room.

"What makes me feel most powerless is not the students' problems, but my own experience. To prepare advanced content for 'Computer Networks,' I used GPT-5 to deeply query frontier ideas for protocol optimization. The summary it provided, the hypotheses it proposed, and even the papers it cited—the breadth and depth exceeded the lesson plan I had prepared for a whole week. Many of us teachers... might be facing a challenge to our knowledge authority."

The conference room was silent.

Finally, the dean took a deep breath.

"Teaching reform must start immediately. We need to pivot to software design, system architecture, ethical review, and human-machine collaboration under AI empowerment."

"But how exactly? Textbooks, syllabi, teacher training... everything is at zero."

Associate Professor Li voiced everyone's confusion.

"We ourselves are still explorers on this new road."

The discussion on the internet was even more chaotic, with anxiety spreading across every platform.

On a forum, a post titled 'Personal experience from a fifth-year NLP graduate student, despair, a true dimensionality reduction attack' was upvoted to god-tier status.

The original poster wrote:

"My advisor is a well-known expert in the field. Our lab spent three years sharpening a sword, improving a specific text generation evaluation metric by 0.5%, and we were preparing to submit to a top conference. I ran the same task using GPT-5, and without any domain-specific fine-tuning, all metrics completely crushed our model, with an improvement of over 10%... That feeling is like practicing archery for ten years and finally being able to hit a target from a hundred paces, only to look up and find your opponent has arrived in a starship."

The reply section turned into a massive scene of resonance:

"So real... the senior in our lab is almost depressed. The database we worked so hard to accumulate is not as good as the zero-shot effect of their general model."

"5-10 year gap? That's optimistic! In terms of engineering and ecosystem, it feels like a generational gap!"

"Basic research is the root!"

"What's the use of the root? Professional titles depend on papers, papers get crushed; companies want implementation, implementation gets crushed..."

"Concentrate efforts to do big things? The problem now is where to concentrate efforts? Direction is ten thousand times more important than effort!"

In a WeChat group called 'Coder Survival Guide,' messages were flying by:

@ Product Manager Old Wang: "Emergency meeting notice: The company has decided to fully implement AI-assisted development, aiming for a 50% increase in human efficiency before Q4. Everyone... good luck."

@ Frontend Little Li: "We are already using it... writing pages now is like making PPTs. Input requirements, AI generates the draft, I fine-tune. The speed is N times faster, but I feel like I'm about to change from an engineer to an AI aesthetic proofreader."

@ Tester Little Zhang: "AI automated test case generation + execution, full coverage + exception injection, it's faster and broader than our manual testing. Our group is likely going to be merged."

@ Architect Old Chen: "Stay calm, everyone! AI lacks business insight and strategic planning capabilities!"

@ Product Manager Old Wang replied to @ Architect Old Chen: "Old Chen, don't lie to yourself. I had it analyze our user behavior data from the past three years. The product iteration suggestions and feature priorities it gave were logically rigorous, fully supported by data, and sharper than the report from the consulting firm we paid a lot of money to hire last time..."

@ UI Designer Little Liu: "I just tried using AI to make a logo and a set of UI, and the director directly said it was more advanced and unified than our current version... I..."

@ HR Little Mei: "From the recruitment side, the most direct impact is the freezing of headcounts for junior technical positions, with social recruitment fully leaning towards senior experts and high-potential talent. In the future, the threshold for newcomers to enter the industry will get higher and higher."

The direction of public opinion is already very clear. From professionals to the general public, from academia to industry, almost everyone is discussing one topic.

How big is the gap we have in the AI field?

And behind this discussion is a deep sense of technological anxiety and uncertainty about the future.