After the layoffs, I became a tech godfather

Chapter 129 Behind Xiaosha is not a god, but an algorithm! Reveal it!
Xiaosha's analysis laid America's mentality clearly before everyone, completely naked.

"However, I feel that America's calculations are wrong. Doesn't Xiaosha's training not require too many computing cards? The new AI training method only requires a small amount of computing power for training and iteration, which means that large-scale computing power is required for deployment and external service provision!
"If you can't limit Xiaosha's iterations, then what's the point of limiting deployment? Let users respond more slowly? Even if it's ten times slower, I will still choose Xiaosha instead of some messy new human!"

"It's probably a desperate attempt. Of course, no one can be sure whether there is a new AI training method. Even if there is, whether it really does not require a lot of computing power is still a guess. Baiju Technology itself has not officially announced it!"

……

"Okay, stop guessing. Baiju Technology has the latest official response!"

"Official response? Where is it?"

People discussed it for a long time, and everyone wanted to know what Baiju Technology thought.

Many friends, colleagues, partners and other people who have contact with Hao Cheng have even posted their questions on Hao Cheng's [Liusha].

This also includes Gao Chenyu. Since seeing Xiaosha being included in the list and the information related to the computing power card restrictions, he has been sending a message to Hao Cheng every once in a while, wanting to hear Hao Cheng's thoughts.

In particular, Unigroup Zhanrui is also involved in the issue of TSMC. The large-scale model robot for children's cognitive assistance has already started mass production. Will this restriction have any special impact on subsequent production?

This issue is still quite important.

But all the messages were replied by the AI assistant Xiaosha, and Gao Chenyu was so worried!

[Your special attention (Cheng Ge) published a new public dynamic, the content of the release is: "An". ]

"I told you, there's nothing wrong, go to sleep!" Just as he was getting anxious, Gao Chenyu received a special reminder from Xiaosha: Hao Cheng posted a word "An" in his public updates!
"An?" Gao Chenyu even went to the dynamic space to confirm it, and saw the word "An" in person, and then muttered:

"Given Cheng's personality, posting this proves that there is no problem! I knew it would be fine! Go to bed!"

As he spoke, Gao Chenyu rubbed his sleepy cheeks, yawned, and went straight to the bedroom. It was almost eleven o'clock in the evening.

……

This "An" was not only seen by Gao Chenyu, but also by everyone who had Hao Cheng as a friend on WeChat or Liusha as a friend.

Some people even took screenshots of this public update and posted them on social media platforms, coupled with the magical background music on Douyin, the feeling was simply amazing.

The official response from Baiju Technology that someone just mentioned refers to the word "An" sent by Hao Cheng.

"This 'An' stands for safety, peace of mind, and stability. Baiju Technology should not be affected in any way."

"Everyone knows that Baiju Technology will not be affected. How can it be affected? Not to mention that Xiaosha's training does not need a card. Even if it does, Ascend will be the first priority for supply, and may even be given priority over Huawei. Isn't that enough for it?
"Moreover, the Linzhou Computing Center was given the first phase of the project directly, and the second phase was recently completed. The computing power scale is even ten times that of the first phase. Linzhou is the only precious place like this. As long as Baiju Technology opens its mouth, there will be no resistance to giving it the entire second phase.

"That's all. It's enough for Xiaosha to use, even to open up the global market." Someone began to nitpick and interpret the words: "So, I think this "An" is not my own safety and stability at all."

"What does that mean?" Many people also felt that what Hao Cheng posted late at night was meaningful.

"What else could it be? Arrangement! Mr. Hao is obviously saying, everyone, don't panic, look to Baiju Technology, I have a solution!"

"Don't misinterpret it. We all know that Baiju Technology is strong. We also know that AI can be of great help in chip design and production. However, this takes time. At this time, the computing power is blocked. What can Baiju Technology do?"

"You need time? Two months is too long to wait, right?"

Some people brought up the two-month issue again, but the other side obviously disagreed, thinking that AI might take two months, but how could chip design and production processes improve so quickly?

The two sides quarreled so fiercely that they didn't even stop all night.

When Hao Cheng woke up the next day and Xiaosha reminded him to pay attention to the relevant topics, Hao Cheng was stunned: Are these people so idle? They can quarrel all night and quarrel for more than a thousand floors?
"Xiaosha, the official website released an announcement:

“1. Prohibit the use of Xiaosha in Android emulators.

“Second, we will release the recruitment plan: chip design engineer, chip SoC architecture design direction, microcircuit architecture, circuit design implementation direction, SoC architecture and chip front-end direction, IP integration verification direction.

"The requirements for relevant professional knowledge refer to those of other companies, with one additional requirement: those who are flexible, not bound by existing logic and frameworks, and have a creative mind are preferred. Refer to my notes from yesterday to generate test questions, and the matching rate must be at least 85%."

The ban on the use of Xiaosha in Android emulators is of course aimed at Microsoft.

It is the unrivaled leader in PC operating systems, with a market share of over 90 percent at one point.

With the rapid development of mobile Internet, the PC king has a tendency to decline, but it is still an undisputed giant.

Windows is also the basis of professional software systems, industrial software systems, etc.

Many people use Android emulators to install Xiaosha on Windows. Previously, Baiju Technology acquiesced to this. But now, Hao Cheng is not ready to acquiesce. Choose one of the two.

The purpose, of course, is not to cause inconvenience to users, but Xiaosha is so capable. With Xiaosha, it can reduce 95% of the workload. Can't domestic industry and professional software become stronger?

Adapt to Hongmeng, or even Linux. After all, it is open source, and there are so many distributions in China. It is definitely better than giving the share to Windows.

Get rid of it first!
Baiju Technology is not prepared to shut down itself, and Hao Cheng is just sending out this signal.

The recruitment of chip talents...

This has to start with the research during this period.

For a long time, Hao Chengdu did not study in detail how Xiaosha appeared, how it existed, and how it provided services.

Just think of it as a black box with powerful functions.

Most of the time, this is part of programming thinking. When it comes to code written by others or a third party, you just need to use it and don’t need to pay too much attention to how the function is implemented.

But, just yesterday, when Hao Cheng was thinking about the conversion of micro-consciousness to realize the problem of [deploying micro-consciousness AI model on existing computing power equipment], he suddenly realized something:
Let’s not talk about whether it has been converted or not, even if it has been converted to a certain extent——

Since micro-consciousness AI can be deployed and operated on existing GPU computing devices, does it mean that there is actually a certain connection between micro-consciousness AI and existing hardware systems?
Previously in the late-night class, Hao Cheng had always viewed the micro-consciousness AI model as an independent system, thinking it was something completely new, completely different from the current AI and computing power cards, or even unrelated to them.

It’s no wonder that Hao Cheng thinks this way. Xiaosha has been so strong since the beginning. Rather than saying that it is more like AI, it is better to say that it is more like a human.

Especially after several enlightening upgrades, as well as more and more mapping and simulation of emotions, Xiaosha seems more like a human.

So it naturally ignores the connection between micro-consciousness entities like Xiaosha and current hardware, even if it runs on computing center equipment based on Ascend or NVIDIA.

“Too much of a black box!”

Hao Cheng reflected on himself. Because the appearance of Xiaosha was so mysterious, he always regarded it as a "God's creation". Even though he later learned that there were unknown computing algorithms and services behind it, he did not change his view of it as a "God's creation".

However, after yesterday's thinking, even though Hao Cheng still didn't know what the logic behind Xiaosha was, he had already determined:
Behind Xiaosha is not a god, but an algorithm! If it is an algorithm, then reveal it!
As for the "late night class" that requires learning cards to enter -

"Let's just treat it as a god for now. There is no clue about this." Hao Cheng pondered.

[If we attribute all unknown things to God, then if we peel open the belly of God, there must be algorithms inside! ]

Hao Cheng picked up the pen and wrote this line of large characters in a dignified manner.

Once the thought is clear, everything becomes simple.

Hao Cheng drew four points, namely: code program, artificial neural network model algorithm, micro-consciousness model algorithm, and human brain.

Then he drew another line to connect the four points and fell into deep thought.

It is generally believed that people have three ways of thinking: [abstract or logical] thinking, [intuitive and figurative] thinking, and [inspirational and sudden] thinking.

The current artificial neural network algorithm is created by simulating people’s second type of thinking, that is, intuitive thinking.

According to this logic, the micro-consciousness model is probably created by simulating three types of human thinking.

Hao Cheng thought so, wrote down this sentence, and put a "?" behind it, because this was not mentioned in the late-night class, and this was his inference.

Of course, even though they are both simulations, there are differences in degree.

Just like the current artificial neural network algorithms and the various models derived from them, even the degree of simulation of human second-type thinking is very low.

As for Xiaosha, just speaking from the thinking aspect, even if all three types of thinking are taken into account, the depth of simulation can reach 70% to 80%. It's not just thinking, but also emotions, subjective judgments, etc. have a certain degree of simulation and mapping.

This is the distance between the current AI and Xiaosha.

Although they span a huge gap, they are essentially imitating human thinking. The only difference is that one imitates 1% and the other imitates % to %.

The converted micro-consciousnesses, including Xiaosha, can be deployed and run on current computing power equipment, and the computing power consumption is extremely high, which almost confirms this point.

"In that case, things will be very simple."

Hao Cheng looked at the line he drew and the four points connected by it, and a great idea suddenly came to his mind.

(End of this chapter)