Rebirth 08: Rise from copycat phones

Chapter 666 Google's Breakthrough in Artificial Intelligence Technology

Late May, Silicon Valley.

At the AI ​​R&D center that Google has built in the last two years, tech talents from all over the world are busy with their own work.

However, most of these technical talents are merely worker ants, unable to work on the underlying core algorithms of artificial intelligence technology.

This seemingly massive Google AI R&D center, with thousands of employees, actually has a core algorithm team of just over ten people, each of whom earns an annual salary of tens of millions of dollars.

These top-tier AI algorithm engineers are that expensive... ever since Zhiyun Group released its AI technology Yun AI early on, and launched AI technology application products such as AI robots and virtual devices.

The global wave of artificial intelligence has also reached a new peak.

Every major high-tech company in the world is working on artificial intelligence technology, especially Google and Microsoft, which are directly developing the core algorithms of artificial intelligence, rather than just advancing the technology based on the open-source algorithms released for free by Zhiyun Group.

They were well aware that while the free and open-source underlying algorithms released by Zhiyun Group seemed advanced, they were actually worlds apart from the algorithms used by Zhiyun Group itself... Zhiyun Group's management wasn't stupid; why would they release their own advanced algorithms?

In reality, the free algorithms released by Zhiyun Group are just above the level of other companies' self-developed algorithms; they are better, but not by much.

It can be used by various businesses, and it works very well in the general business sector.

The reason so many companies are deploying GTAI for various artificial intelligence applications is not simply because Zhiyun Group has a big reputation, but because GTAI is genuinely easy to use.

However... you can't expect to surpass Zhiyun Group using a free algorithm like GTAI!
Google is well aware of this, so although they also use GTAI and other free algorithms published by Zhiyun Group for commercial applications, they have never given up on developing their own algorithms.

In fact, Google's investment in self-developed algorithms is increasing every year.

Similar situations exist with companies like Microsoft, Tesla, and even Apple; they have not actually given up on developing their own underlying algorithms for artificial intelligence!
Even among a host of high-tech companies in China, apart from Yihai Technology, which is also owned by Xu Shenxue, other top high-tech companies have not given up on developing their own algorithms. Companies like Huawei, Tencent, and Alibaba are all developing their own algorithms.

The influx of companies into artificial intelligence technology has led to a surge in the net worth of key technical personnel in the field, with their annual salaries becoming increasingly extravagant year by year.

Of course, this mainly refers to core technical talents such as algorithm engineers, and does not include some auxiliary personnel who seem to work in the artificial intelligence industry but are actually highly replaceable.

For example, some people label algorithms or correct and train algorithms. The salary for this kind of job in China is only a few thousand yuan...

A typical example is the intelligent robot project under the Zhiyun Group, which has a large number of outsourced training factories.

Training workers in outsourced training factories train and correct specific functional models of robots day and night.

And the wages for these training workers are only a few thousand yuan...

Despite the current boom in artificial intelligence technology, the industry with the highest median salary is actually not artificial intelligence, but the virtual content industry.

However, this does not mean that technical personnel working in artificial intelligence have low incomes; rather, it means that the compensation levels in the artificial intelligence industry are severely polarized.

In the artificial intelligence industry, the vast majority of practitioners are actually assistants in the training field, such as labeling algorithms and training them to perform actions.

You might say they aren't AI professionals, and they are indeed in the AI ​​industry, but their work has little to do with technology. They can basically be seen as assembly line workers and ordinary white-collar workers in cheap offices.

These people are highly replaceable and don't need to do any research and development work, so their income is generally quite average.

However... the technical talents who develop artificial intelligence earn much more, with a median annual salary of around 560,000.

For algorithm engineers in core positions, an annual salary of several hundred thousand is already considered a beginner's salary. Those with a little experience and better abilities are basically earning millions of yuan a year.

These core technical personnel often earn annual salaries of several million or even tens of millions of yuan.

As for top talents who can take on major responsibilities, solve underlying core algorithm technology and theoretical problems, and develop advanced underlying algorithms, their annual salary is tens of millions or even hundreds of millions.

Why is the AI ​​Research Institute of Zhiyun Group so amazing? Because they have a group of top AI talents with annual salaries starting at tens of millions, or even hundreds of millions of yuan.

It's not just one or two, but a whole group... Each of the algorithm engineers in them is enough to make companies like Google drool.

Google has searched the entire world for such top talents and only managed to form a small team... while Zhiyun Group has a whole bunch of them!

This gap left even the senior management of Google feeling somewhat hopeless... The number of top talents they had wasn't even a fraction of what the other company had!
This situation is unlikely to change in the short term... because in the past decade, the place that has cultivated top AI talent globally has not been the United States, but China.

Globally, at least 80 of the top 100 AI algorithm experts were trained by universities in China, while the remaining 20 were trained by universities around the world.

Therefore, even among the AI ​​technology talents employed by various American companies today, the majority are actually of Chinese descent...

This makes it much more difficult for Chinese companies like Google and Zhiyun Group to compete for top AI talent!

After graduation, these top talents often choose to go directly to Zhiyun Group rather than go to work for American companies.

For the past seven or eight years, the best job option for graduates from China's top 985 universities has been to join Zhiyun Group during campus recruitment.

Once you get in, you can get an extremely high salary that's unmatched in the country... Earning a million a year is highly likely, and you don't have to worry about being laid off in middle age for no reason.

You can also gain a respectable social status. When you go on a blind date, you can say that you are an employee of Zhiyun Group, and your date will look at you with more respect... A famous saying in the blind date market: Zhiyun Select!
There's also a world-class stage... If you want to do scientific research, as long as you're incredibly capable, you can squander the funding as much as you want.

If you take the administrative management route, the day you reach a senior management level, local leaders will pick you up at the airport when you go abroad for business trips.

If you take the financial route, you'll be involved in the operation of funds worth hundreds of billions or even trillions. At that level, you can decide on projects worth tens or hundreds of billions with just a word.

If we're talking about marketing and public relations, let's put marketing aside for now and talk about the public relations approach first... The senior public relations staff at Zhiyun Group don't just drink with clients; they directly liaise with high-level officials from various countries and engage in political lobbying.

If you are good enough and excel in your field, you will have a world-class stage to showcase your talents.

Generous compensation, a high social standing, and a world-class platform are all major attractions for top domestic talent at Zhiyun Group.

People only consider other options when they can't get into Zhiyun Group.

Examples of career options include: being selected for a government position; joining a top-tier domestic company other than Zhiyun Group through campus recruitment; pursuing a master's or doctoral degree in China; pursuing a master's or doctoral degree at a prestigious overseas university; or joining other medium-sized or large enterprises.

In short, for the vast majority of top talents, Zhiyun is the ultimate dream, and everything else is just a backup plan...

Moreover, the above choices apply to both arts and science students!
After all, Zhiyun Group is a large company with all sorts of diverse businesses, so no matter what major a student is in, Zhiyun Group will recruit them, and once they join, they will have a platform where they can showcase their talents.

But... they only want the very top-tier batch!
To put it bluntly, top domestic talent is often first selected by Zhiyun Group through a competitive process, and only after this process is completed can other industries or companies begin to recruit. This makes it quite difficult for overseas companies like Google to recruit AI talent.

On the one hand, it attracts top talents from other countries, and on the other hand, it tries every means to recruit top talents in China, even at the cost of higher salaries.

Even so, the quality of the artificial intelligence technology team that Google has assembled is not very good.

Compared to other companies, they seem quite impressive, but unfortunately, their benchmark is Zhuyun Group...

------

Today, the entire algorithm team at Google is in their office, staring intently at a large display screen in front of them.

"The results are almost in; in five minutes at most, we'll know exactly how our new algorithm performs!"

"The completely new algorithm architecture, including the underlying mathematical tools, has been significantly replaced and upgraded. At least theoretically, our new algorithm should be at least twice as efficient in training as the previous generation!"

"Furthermore, the need for algorithms will be further reduced!"

"Our new algorithm has been specifically optimized and modified for the characteristics of superconducting quantum computers, and can fully utilize the computing power of superconducting quantum computers!"

"Even if its performance isn't as good as Yun AI, it shouldn't be too far off!"

Five minutes passed quickly. After the time was up, everyone looked at the results on the monitor, and smiles immediately spread across their faces.

"We succeeded! We succeeded! Look, our new algorithm is so well adapted to quantum computing! It's at least twenty times more adaptable than our old algorithm!"

"The algorithm is using too much computing power, which is beyond our expectations!"

"The output speed is extremely fast!"

"Most importantly, it has a reasoning process. Look at how rigorous this reasoning process is!"

"It's not just rigorous; it also fills in several possibilities that weren't explicitly mentioned in the question, conducts multi-path deductions, and after multiple rounds of comparison, selects the correct result!"

"Based on these results, the large-scale generative model using the new underlying algorithm, while still performing slightly worse than Yun AI, should be able to achieve about 80% of the performance!"

"Our new algorithm is a success!"

Google's top management quickly learned that their artificial intelligence research center had achieved a major technological breakthrough... not that a certain big data model had been successfully trained, but that a technological breakthrough had been achieved in the field of underlying core algorithms.

Before long, several senior executives from Google rushed over to take a look.

Before long, Google made a decision to further increase its investment in the field of artificial intelligence, purchase more superconducting quantum computers, and build more computing centers.

They want to take advantage of the breakthrough in the underlying algorithm technology to quickly continue to break through along this technological path and finally apply it to other applications, such as creating better large-scale generative AI.

At the same time, we need to explore whether this new breakthrough in technology can be applied to the field of intelligent robots, such as training an artificial intelligence big data model for an intelligent robot.

The breakthrough in underlying technology has greatly excited the top management of Google... even though technical experts told them that their technological breakthroughs are still a long way from the artificial intelligence technology of Zhiyun Group, and even inferior to the artificial intelligence technology of Hailan Automotive Group.

However, this did not dampen the excitement of Google's top management.

As the technical experts have also said, although there is still a technological gap, and it is quite large... their new algorithm is theoretically sufficient to support the underlying algorithm requirements of intelligent robots and virtual devices.

The intelligent robots made using this algorithm are naturally far behind the intelligent robots from Zhiyun Group in terms of performance, but at least they are usable.

Unlike those previous garbage algorithms that produced intelligent robots that were idiotic and would fall over after taking just a couple of steps.

Next, the algorithm needs continuous optimization, and some specific application-oriented big data models need to be trained based on this new algorithm theory...

This requires a massive AI computing center!

It's not a traditional GPU computing center, but a new type of quantum computer computing center... because their new algorithms are designed based on the characteristics of quantum computing.

Without the massive computing power provided by quantum computers, it's simply impossible to do!
Without hesitation, the executives at Gugou placed a new order with Zhiyun Group shortly afterward!

The order amount is quite staggering, exceeding a staggering $100 billion!

Google not only purchased a large number of YCL-52 superconducting quantum computers, but also pre-ordered a batch of second-generation GTL servers, namely GTL2 servers, to support them.

This GTL2 server uses an APO7000 graphics card.

Although the APO7000 graphics card has not been officially released to the public, as a computing power card that is used internally and sold externally, Zhiyun Group has not kept it a secret from strategic customers such as Google.

We have also notified strategic customers such as Google that the APO7000 graphics card, which uses an equivalent 3-nanometer process and fourth-generation high-bandwidth memory, has been successfully taped out and is expected to start mass production in December, with large-scale supply expected around January next year.

The GTL2 server not only uses the APO7000 graphics card, but also the more advanced WZ400 server CPU manufactured using a 3-nanometer process, and the most advanced high-speed enterprise-grade memory using the 10DE process.

For these core components, overseas semiconductor manufacturers such as Intel and AMD also have no substitutes... The technology is too advanced, and they are struggling to keep up.

Google, on the other hand, simply stopped insisting on using the chips from these companies and instead planned to directly purchase a complete set of GTL2 servers with Zhiyun chips.

They need the most advanced computing systems right now, and in large quantities.

New algorithms have been developed, and all that's left is to have the computing power so that they can train large-scale application models. Google is getting impatient too!

There are various artificial intelligence application markets waiting for them, especially the two emerging industries of robotics and virtual equipment.

They no longer care about the fate of Intel and AMD...

As for the bottleneck issue... instead of worrying about being held back by computing power bottlenecks, we should worry about Zhiyun cutting off the supply of superconducting quantum computers.

GPU computing cards and advanced server CPU chips are actually quite good. They have some of the less powerful ones, and they can still be used without major problems.

However, when it comes to superconducting quantum computers, there's really no way to get one; they don't even have a low-end alternative.

(End of this chapter)