Tutor's Complaints

"The school's funding is a bit insufficient. I'll ask Li Youcai if it's possible to acquire some equipment."

After hearing Professor Zhang's words, Lin Yuan's mind finally felt a little more at ease.

It seemed that even though his mentor also thought his idea of "algorithms producing algorithms" was impractical, he was still willing to support his endeavors.

As for why Lin Yuan had to stubbornly make things up, it was because he couldn't very well say, "The system taught me everything."

The computing power system could actively iterate on the AI model itself during AI model training, much like a supercomputer with a brain. However, this capability was beyond the current level of human technological development. Otherwise, Lin Yuan wouldn't be considered a cheat.

Lin Yuan's biggest problem wasn't actually the system's lack of computing power. It was how to find a perfect explanation to prevent people from becoming suspicious when he applied the system's capabilities to reality.

You should know that in engineering and science, there's no shortage of straightforward, no-nonsense people.

In novels, one can write, "A melancholic cloud drifted across the sky," and liberal arts students wouldn't be too fixated on it. But among those who study engineering and science, someone will always ask, "How do you prove that cloud is melancholic?"

What is your theoretical basis? Do you have experimental data...?

Lin Yuan pondered this question for a long time. Finally, he found inspiration from the characteristics of the computing power system itself.

What is the system?

If the system itself is an overpowered AI, then the process of the system running AI training isn't just AI training AI, is it?

Thus, Lin Yuan connected (and forced) the idea to—algorithms generating algorithms.

This viewpoint, well, you certainly can't say it's wrong; it even has some merit.

It's like someone proposing the idea: If I could calculate every quark in the universe, I would know the entire past and future of the universe.

How do you prove that statement is wrong? Because you simply can't truly calculate every quark in the universe. And the statement seems quite logical, even possessing the wild imagination of science fiction and the ethereal beauty of philosophy.

So, for algorithms to produce algorithms, how do you prove that this won't work?

Even if I can't prove it works now, even if I fail in the future, my failure doesn't prove that my theory is wrong; it only proves that I can't prove the theory I proposed.

However, Lin Yuan kept these forced ideas to himself. He wouldn't dare say them aloud, fearing his mentor might strike him down with a single blow.

Perhaps Lin Yuan's honest and tenacious personality made Professor Zhang view his almost fallacious idea as a stubborn persistence, like a bull. And when a person reaches a certain age, they might find every bull endearing.

Because that bull might have been their past self.

"This Li Youcai just knows how to complain about being poor." Professor Zhang slammed his phone on the table, pulling Lin Yuan's thoughts back.

Just now, Professor Zhang had spoken with Dean Li, and the other party clearly did not agree with the idea of acquiring more computing equipment.

"Teacher, if it's inconvenient, then forget it. I'll find another way myself." Lin Yuan didn't want his own matters to cause other issues.

"It's not just for you. The institute's equipment is already lacking; it should have been purchased long ago. That turtle son Li Youcai only cares about his political achievements and the number of papers published. What's the point of throwing so much money into publishing papers every year? So many SCI papers each year, yet I haven't seen any economic output. They squander their inheritance without a care, a bunch of prodigal sons.

Was Musk's Starship built on a pile of SCI papers? They're putting the cart before the horse, reversing cause and effect, it's utter nonsense..."

Professor Zhang went on a furious rant, even getting so worked up that he spoke without a single cough.

"Teacher, please calm down quickly." Lin Yuan hadn't expected his mentor's firepower to be so fierce. He felt that if he didn't interrupt him soon, he might start spouting even more explosive content.

"I can ask the senior students to borrow more mining cards. Then, we can find a distributed AI training framework; working together might also work."

So-called distributed AI training is breaking down a large task and assigning it to different machines. However, just like human division of labor, machine division of labor also incurs management costs. Even if Lin Yuan could borrow a bunch of mining cards, getting them to cooperate and run normally wouldn't be a simple matter.

What's more, Lin Yuan probably couldn't borrow a bunch of mining cards anyway.

"Mining cards?" Professor Zhang's previously calmed emotions were ignited again. "The money spent on each graduate student for papers is more than the money for buying equipment. Then you kids are made to work on topics that are either detached from reality or have already been overdone by companies.

Take image recognition, for example. Things like OCR character recognition have already been overdone in companies; what's there to research?

They spend all day researching how to make light bulbs. They call it: mastering the underlying principles.

How many Einsteins does humanity produce in a hundred years? Are people of Einstein's caliber cultivated?

They don't base themselves on societal needs or practical applications. That's why, in the AI field today, universities' research achievements can no longer keep up with companies.

Was the Double 11 recommendation algorithm developed in some university laboratory?

If you say this to their faces, they all act arrogant and say it's just engineering application.

Then you ask them about their research achievements, and they show you data. The data looks beautiful. But look at the influential papers in the AI field over the past twenty years; which one did they produce?"

Lin Yuan stood by, listening to his mentor complain, completely unsure how to respond.

After a while, Professor Zhang finally vented his frustrations.

"Ah, I'm getting old; I tend to speak without thinking.

What I just said isn't entirely correct; pick and choose what you listen to. After all, you still need to prioritize graduating smoothly."

Lin Yuan naturally understood that if he had even one-tenth of the rebellious spirit in his mentor's words just now, he would already be finished.

Lin Yuan also understood that his mentor was using the criticism of his own university to implicitly criticize the current state of scientific research in China. His mentor didn't harbor the common public intellectual views often seen online, such as criticizing the domestic university education system or the college entrance examination system. Instead, he was complaining about the issue of disconnect between universities and enterprises.

And judging from his mentor's words, he clearly placed more importance on engineering applications. This happened to align with Lin Yuan's interests.

In choosing between academic and applied interests, Lin Yuan seemed to have never hesitated; he seemed to have gone straight for engineering applications from the beginning. This was probably also related to him taking the postgraduate entrance exam because he couldn't bear the hardship of delivering food.

However, after his mentor rambled on for quite a while, Lin Yuan still didn't see any hope for solving the computing equipment problem.

So what was the point of his mentor telling him to return and avoid Gu Xuan? It couldn't just be to watch his old man's angry outburst, could it?

"It's all because Li Youcai and the others forced me, MLG forced me. I still have to go back to my old ways... I'm already this old and still have to do such things."

His mentor's colorful language made Lin Yuan's ears wince, but he keenly caught on, "Teacher, what 'old ways'?"