Birth (Part One)

As an IT professional who keeps up with the latest information, Xiang Kun naturally knows what the most cutting-edge and advanced AI looks like.

In the field of AI, deep learning and neural networks still dominate. Domestically and internationally, major companies such as Google, Tencent, Baidu, and Alibaba all have their own areas of in-depth research and frequently release updates on their latest progress.

After all, most current AI research directions are difficult to develop successfully in isolation. This is especially true for machine learning, where data, computing power, and algorithms are all indispensable. Without sufficient datasets, no matter how sophisticated the algorithm or how powerful the machine's computing power, it is difficult to obtain the desired results and further apply them in real-world scenarios.

In certain specialized fields, AI has already surpassed or even far outpaced human capabilities.

Just as AlphaGo has repeatedly defeated the top Go players, with further improvements and iterations, it may become very difficult for human players to compete with it in the future, and its opponents will become other AIs.

But in reality, was it truly only AI, only machines, that defeated human players like Ke Jie?

That's probably not the right way to put it.

Just like Xiaoming, whose combat power is only 5, he wants to see his lover Xiaohong, but he is blocked by a giant dragon with a combat power of 100 on the way, and he can't beat it.

So Xiaoming returned to his village, gathered his dwarf friends, and spent several years conducting various simulated battles against the dragon, targeting its weaknesses and fighting styles. They then forged a flying sword called the "White Emperor Holy Sword."

The "White Emperor's Holy Sword" has a combat power of 200 when fighting dragons!

So Xiaoming, shouting "White Emperor's Holy Sword, follow me!", easily killed the dragon and met Xiaohong.

So, was the dragon defeated by the "White Emperor's Holy Sword," or by Xiaoming and his dwarf companions?

Without Xiaoming, the "White Emperor Holy Sword" might have been easily defeated by a cat with a combat power of 3, because it had never seen one in its training manual!

Last year, the BERT model from Google even achieved top scores on 11 NLP tasks in SQuAD 1.1, a top-level test for machine reading comprehension, claiming to far surpass humans in reading comprehension. Subsequently, domestic teams from iFlytek, Tencent, and Alibaba also took turns topping the SQuAD test.

With the advent of SQuAD 2.0, teams from Microsoft Asia Research Institute and Alibaba have been continuously setting new and better results, suggesting that in terms of reading comprehension, humans have been left far behind by AI.

But is that really the case? Has AI truly developed a stronger reading comprehension ability than humans? Has it already been able to understand human thoughts and ideas?

Ultimately, current AI is still just an "advanced tool".

No matter how much it looks like a human.

When Xiang Kun first started the "AI Assistant Project," he already saw the bottlenecks in machine learning and neural networks. So initially, he didn't have the idea of creating a "super AI" because that was unrealistic. He was always thinking about integrating and optimizing existing frameworks.

But he later discovered that that approach still required a significant amount of time and effort, and was still not something he could easily accomplish on his own. Since it required so much effort, why not try to take it a step further?

To put it bluntly, after several mutations, he became arrogant again, wanting to eat the best wontons in the world with just two cents in his hand.

He felt that his two cents were very valuable, and it would be a waste to buy something worth two cents, two dollars, or even twenty dollars.

But how do we do it?

Xiang Kun had countless moments of inspiration, vaguely forming an idea, but he could never quite grasp it completely, and it still couldn't fully manifest itself.

This is why he agreed to take on that project for Zhang Qian; he wanted to help her find "that light" again while working on other projects.

Unfortunately, it didn't help much. The framework of that project was already in place and the core was complete, but Xiang Kun's "AI Assistant Project" still hadn't even found the right angle to dig a hole.

Looking at the diagram he had drawn on the screen to help him think, Xiang Kun pinched his chin with one hand and played with the Arthas he had sculpted himself with the other.

Xiang Kun now understands:

To create something worth two million with just two cents, there must be a groundbreaking innovation, something entirely new starting from the ground up, a "lever" capable of moving the earth.

Within the existing framework, he definitely couldn't have done it. Even if, through mutation, his brain was many times stronger than that of ordinary people or even ordinary geniuses in certain areas, he still couldn't compete with so many institutions, large factories, talents, and teams around the world gathered together to conduct research every day, with so much equipment and so much data available to them. Since they couldn't find a breakthrough within the existing framework, it would be equally difficult for him to do it alone.

But how deep does this "bottom layer" have to be?

In truth, software innovation is always inseparable from hardware.

Just like when computers were first invented, artificial intelligence was divided into "symbolicism" and "connectionism," as well as the later "behavioralism."

"Connectivityism," also known as bionics, advocates mimicking human neurons and using the connection mechanisms of neural networks to achieve artificial intelligence. The currently popular machine learning follows the principles of "connectivityism."

However, in the 1970s, due to the level of hardware development and the lack of computing power, "connectivity" once fell into a trough.

Until now, computing power has advanced by leaps and bounds. The development of the Internet and mobile networks has brought about massive amounts of big data, which has provided the most suitable environment for machine learning under the "connectivity" approach to grow. It has begun to flourish and has given rise to a variety of platforms, algorithms and models. They have begun to be widely used in various industries and environments, and there are even claims that "AI will replace human jobs".

Earlier this year, Huawei released the Ascentage 910 AI training chip and the Ascentage 310 AI inference chip. Not long after, Alibaba's DAMO Academy also released the Hanguang 800 AI inference chip, which is touted as the world's most powerful. These hardware improvements will undoubtedly enable machine learning to evolve to a more extreme level.

In the foreseeable future, perhaps in thirty or twenty years, a major breakthrough will be achieved in the field of quantum computing, and quantum computers will begin to be used on a large scale. At that time, it will inevitably drive a revolution in the field of AI, and more comprehensive, complex and powerful AI models will inevitably emerge. Perhaps with the advancement of research on human neurons, a more comprehensive neural network will appear, thus realizing a true "super AI".

But Xiang Kun couldn't wait that long. He didn't have enough money, time, or the will to invest in hardware development and research. He wasn't trying to lay the foundation for the future; he just needed an "AI assistant" that could help him now—and one that could actually be helpful.

Therefore, his "innovation" and his "foundation" can only be built on existing hardware conditions.

It sounds like wishful thinking, but Xiang Kun really felt like he had grasped the key point. It seemed like he was just one membrane away from breaking through, and once he did, everything would become clear.

Xiang Kun checked the time; it wasn't too late. So he picked up his phone and texted WeChat Little Apple, "Any plans for this weekend? Why don't you have your dad bring you over? I'll cook you something delicious, and I'll take 'Gold Glitter' out to play too."

"Okay! But won't this take up too much of your time, Uncle Xiang?" Little Apple asked, her voice clearly a mix of joy and worry.

Xiang Kun replied, "There's nothing to delay. I'm a freelancer now, and I can arrange my own working hours."

After agreeing on a time for them to come over, Xiang Kun casually remarked, "By the way, Little Apple, are you interested in computers?"

"Computers? I only know a little bit. When I was in school for the blind, I took some courses, like the basic structure of computers and stuff, but I don't really understand the specifics," Little Apple said honestly.

"How about I teach you? You know, computers, especially AI, are used in a wide range of fields now. Maybe in the future there will be AI that can interpret 'bird language' and help you understand the habits of various animals. You can learn about it and see if you'll find it interesting," Xiang Kun continued to guide.

"But I have no prior knowledge, so I'd better try to learn on my own first..."

"No need, I can teach you from the beginning. Just ask me anything you don't understand," Xiang Kun said firmly.

Then he began to use simple and easy-to-understand language to popularize computer knowledge for Little Apple, starting from the hardware, including how the CPU works, what the principle is, and how semiconductors are used to form gate circuits.

It must be said that Little Apple is indeed a very studious child. Especially after Xiang Kun mentioned the future applications of computer AI in ornithology and zoology, it piqued her interest and made her willing to learn.

Moreover, Little Apple is quite intelligent. Although Xiang Kun tries to make many of his statements easy to understand, they are still rather abstract since they don't involve real objects. However, Little Apple can understand them very quickly. Even if she doesn't understand something, she can understand it immediately after asking a couple of questions.

Around 11 o'clock, sensing that Little Apple's voice sounded tired and sleepy, Xiang Kun voluntarily ended the "lesson" and said that if there were any questions, she could leave him a message on WeChat and they could continue tomorrow.

After ending his call with Little Apple, Xiang Kun started picking up the pile of boxwood at home—he had gone out last week, and in addition to replenishing the rabbits, he had also found his old supplier and bought a bunch of boxwood of different sizes for backup.

This is where the wood came in handy. Xiang Kun planned to carve it into computer parts and then use the emotional patterns of "interest" and "focus" to "Emotional Injection."

He will complete a blood-drinking period before the weekend, at which point these "Emotional Injection" items will be ready to take effect.

For the past two days, Xiang Kun has been sensing the "Minion Wooden Carving" at Little Apple's place every other day, guiding her to dream, and then "observing" it in her dream.

In the beginning, after he left Little Apple's dream, the only things he could build cognitive models with Little Apple's special cognitive pattern were those that had appeared in Little Apple's dream, or that Little Apple had come into contact with or heard about.

However, after two blood-drinking experiences and two mutations, this cognitive pattern has subtly become Xiang Kun's abilities.

It's like before, you just installed the software, but there was a hardware conflict on your local machine, so the software could only run in the server— Little Apple's dream. Now, after the upgrade, his own "hardware" has also started to adapt.

Of course, compared to Little Apple's cognitive pattern, he still finds it difficult to establish that special cognitive pattern for any object. He can only have a very general feeling, which is mixed with his original cognition and appears somewhat chaotic.

The "flash of inspiration" that Xiang Kun had when he was working on his "AI Assistant Project" came not only from ideas that arose from his own mutation process and mutation model, but also from the special cognitive patterns he gained in Little Apple's dream.

He had a vague idea of combining the two "things," but he couldn't find a good starting point. After all, he hadn't fully figured out what was going on with either his own mutation or the special cognitive pattern from Little Apple; it was sometimes accurate and sometimes not.

So now he wants to guide Little Apple to learn about computers, and then use her special cognitive methods to build a cognitive model of how computers work. Xiang Kun will then use Little Apple's dreams to perceive that model and see if he can gain any inspiration.

Compared to continuing to bury himself in his own thoughts, Xiang Kun felt that this approach might be more worthwhile.

Besides, he really likes Little Apple, and teaching her some computer knowledge might be useful in the future, right? After all, it's the information age.

Over the weekend, Xiao Pingguo enjoyed a delicious meal prepared by Xiang Kun home, played with "Gold Glitter" for a while, and was then enthusiastically tutored by Xiang Kun on computer-related knowledge for several hours. He also gave her a bunch of "audio teaching materials" that he recorded himself. After receiving three very special wood carvings, Little Apple and her father, Li Yang, boarded the bus home.

"Uncle Xiang's cooking is even better than last time!" Little Apple said with a hint of regret, "It's a pity Uncle Xiang doesn't live in the same city as us, otherwise I could mooch meals from him every now and then, hehe."

"I never knew you were such a glutton!" Li Yang said to his daughter with a smile.

"Who doesn't like delicious food!" Little Apple said, carefully touching a wooden box in her hand. "And these wood carvings are amazing too. Those 'capacitors' are so small, the circuits are so fine, yet Uncle Xiang was able to carve out the marks. He's really incredible."

The wooden box Xiang Kun gave her contained three wood carvings of different sizes: a scaled-down version of a motherboard, CPU, and memory module. It wasn't that Xiang Kun was unwilling to carve them to be 1:1 scale, but rather that he couldn't find a piece of wood large enough for the time being.

When Little Apple was at Xiang Kun's house, Xiang Kun specially disassembled a computer he had bought for a project and guided Little Apple to carefully touch the internal components of the computer, such as the motherboard and CPU. He explained to her in detail what each part was, what its function was, and compared it with what he had taught her on WeChat.

Looking at the wooden box in Little Apple's hand, Li Yang sighed, "Xiang Kun is indeed a good person. You can tell he genuinely wants to teach you computer skills."

Thank you to the readers Dark Moon Dawn, DuduDragon, Reader20190926111527226, BookYou1111111111, idateyou, HeartSlightPain, Winlong, and RickSalmon for their donations on Qidian Reading iOS.