What Is The Experience of Becoming a Vampire?

As an IT practitioner, and has been learning about the latest information, Xiang Kun naturally knows what the most cutting-edge and peak AI looks like now.

In the field of AI, it is still the world of deep learning and neural networks. At home and abroad, Brother Gou, Penguin, Du Niang, Ali Niang, and other major companies have their own research directions, and they will publish the latest progress from time to time. .

After all, most of the current research directions of AI are difficult to build good cars by building cars behind closed doors. Especially for machine learning, data, computing power, and algorithms are all indispensable. Without sufficient data sets, you are even more sophisticated in algorithms, and the computing power of machines is NB. It is difficult to get the desired results and further realistic applications.

In some special fields, AI has the ability to surpass or far from human beings.

It’s as if AlphaGo has defeated the top Go masters in succession. With further improvement and iteration, it may become difficult for human chess players to fight against it in the future, and its opponents will become other AIs.

But really speaking of which defeated Ke Jie and the others, is it really just AI, just a machine?

I’m afraid I can’t say the same.

Just like Xiaoming has a battle strength of only 5, he wants to meet his lover Xiaohong, but he is blocked by a dragon with a battle strength of 100 on the road, which he can’t beat.

So Xiaoming returned to the village, gathered his dwarf buddies, spent several years, conducted various simulated confrontations against the dragon’s various weaknesses and fighting methods, and then created a Flying Sword “White Emperor sacred sword”.

“White Emperor Sacred Sword” battle strength 200 when fighting dragons!

So Xiaoming shouting “Baidi sacred sword followed me!” Xiao Ming easily killed the dragon and saw Xiao Hong.

So is the dragon defeated by the “Sacred sword” or is Xiaoming and his dwarf companion?

Without Xiaoming, maybe “White Emperor Sacred Sword” encountered a cat with a battle strength of 3 and was shot down because he hadn’t seen it in his training set ~!

Brother Gou ’s BERT model last year, and even the 1.1 NLP tasks in the SQuAD11 test of machine reading comprehension won STOA results, claiming to be far superior to humans in reading comprehension. Subsequently, the domestic teams of Xunfei Brother, Penguin Brother, and Ali Niang also successively reached the top in the SQuAD test.

At SQuAD2.0, the teams such as Giant Asian Academy and Ali Niang continued to refresh their performances. It seems that humans have also been thrown far away by AI in terms of reading comprehension.

But is that really the case? Is AI really better at reading comprehension than humans? Already understand human thoughts and ideas?

After all, the current AI is still just “advanced tools”.

No matter how much it looks like humans.

When Xiang Kun first started the “AI Assistant Program”, he already saw the current machine learning and neural network bottleneck, so at the beginning, he did not at all to do the “super AI” idea, because that was not realistic, he has been Is to do an integration and optimization under the existing framework.

But later he discovered that that method still took a lot of time and energy, and it was still not something he could easily accomplish by himself. And since it takes so much effort, why don’t you want to go further?

To put it plainly, even after a few mutations, he “swelled” again, holding two cents, and wanted to eat the best ravioli in the world.

Because he feels that these two cents are very valuable, it is a loss to buy something that is worth two cents or even two yuan or 20 yuan.

But what to do?

Xiang Kun has had such a bit of “divine light flashed” countless times, and has vaguely had an idea, but there is still no way to fully sacrifice it, or he can’t come out.

This is why he promised Zhang Qian to take over that project. He wanted to help him find the “light” in the course of other projects.

But unfortunately, the help is not great. The framework of that project has been completed and the core has been completed. However, Xiang Kun’s “AI Assistant Program” still does not find the angle of digging well.

Looking at the picture he drew on the display to help himself think, Xiang Kun pinched his chin with one hand and played with his own carved Alsace.

Xiang Kun now understands:

If you want to make 2,000,000 things with two cents, you must have a very groundbreaking creation. It must be a brand new thing starting from the bottom, and a “lever” that can pry Earth.

Under the old framework, he definitely can’t do it. Even if he mutates, the brain is much stronger than ordinary person or even a general genius in some fields, but the whole world that many institutions, that many big factories That many talents and team come together to study every day. There are many devices and that many data available for them to use. He is incomparable by himself in any case. Since they can’t find a breakthrough point under the old framework, it is also very difficult for him to do it alone.

But how low is this bottom?

Really speaking of which, software-level innovation can never be separated from hardware.

As if when the computer was first born, artificial intelligence was divided into “symbolism” and “connectionism” and later “behaviorism.”

“Connectionism” is also called bionics. It advocates imitating human neurons and using artificial neural networks to implement artificial intelligence. Nowadays, the hottest machine learning practice is “connectionism.”

But in the 70s, because of the level of hardware development and lack of computing power, “connectionism” fell into a trough.

Until now, the computing power of computers has been advanced by leaps and bounds. Because the development of the Internet and mobile networks has brought massive amounts of big data, machine learning under “connectionism” has the most suitable growth environment and has begun to flourish. A variety of platforms, algorithms, and models have been bred, which have begun to have a large number of applications in all walks of life and in various environments, and even the saying that “AI wants to replace human work”.

Earlier this year, Huawei released the Shengteng 910 AI training chip and Shengteng 310 AI inference chip. It didn’t take long for Alibaba to release the so-called world’s strongest light-containing 800AI inference chip. These hardware improvements will be necessary. Will allow machine learning to achieve a more extreme evolution.

For the foreseeable future, perhaps 30 or 20 years, quantum computing has achieved significant breakthroughs. Quantum computers have begun to be applied on a large scale. At that time, it will inevitably lead to changes in the AI ​​field. A more comprehensive, complex and powerful AI will be born. The model, maybe with the advancement of human neuron research, there will be more comprehensive neural networks that “plagiarize”, thus realizing “super AI” in the true sense.

But Xiang Kun ca n’t wait until then, and he does n’t have enough money, time, or willingness to invest in hardware development and research. He does n’t have to be in order to lay the foundation for the future. He just needs an “AI that can help him now. “Assistant”-and really help.

So his “innovation” and his “bottom level” can still only be built on existing hardware conditions.

It sounds like dreaming in the day, but Xiang Kun really feels that he has caught a little tail, it seems that it’s a difference of only one level film, and he can be suddenly cheerful after breaking through.

Xiang Kun watched the time, it wasn’t too late, so he picked up the phone and said to Apple with WeChat: “Are there any plans for this weekend? Let your dad take you to play, I will cook for you, by the way Bring ‘golden glitter’ to play. “

“Okay! But this will not delay your time with Uncle, right?” Little Apple’s voice was obviously a little happy, but asked a little worried.

Xiang Kun replied, “There is nothing to delay. I am a freelancer now, and I arrange my working hours.”

After agreeing on the time for them to come and play, Xiang Kun said unintentionally again, “Yes, little Apple, are you interested in computers?”

“Computer? I only know a little bit. When I was in the blind school, I had taken some courses, that is, the basic structure of the computer and so on. I didn’t understand it specifically,” said Little Apple Frank Frankly.

“How do I teach you? You know, the field of computers, especially AI applications is very wide, and maybe there will be AI for the interpretation of” bird language “in the future, which can help you understand the habits of various animals, you also You can learn and understand, maybe you will be interested? “Xiang Kun continued to guide.

“But I don’t have any foundation, or I’ll teach myself …”

“No, I can teach you from scratch. Ask me anything you don’t understand,” Xiang Kun said firmly.

Then he started to use simple and easy-to-understand language to popularize Apple’s computer knowledge from hardware, even including how the CPU works, what the principle is, and how to use Semiconductor to form each and everyone gate circuit.

I have to say that Little Apple is indeed a very easy to learn child. In particular, Xiang Kun mentioned that the future application of computer AI in ornithology and zoology has aroused her interest to some extent, and made her willing to learn.

Moreover, the IQ of Little Apple is also very high. Although many of Xiang Kun’s statements are as easy to understand as possible, after all, there are no real objects and they are relatively abstract. However, Little Apple can understand quickly, even if it is unclear, Asking two sentences will immediately understand.

As soon as eleven o’clock, after feeling a little tiredness and drowsiness in the voice of Little Apple, Xiang Kun actively ended the “course” and said that if there was any problem, he could just leave a message on WeChat and continue tomorrow.

After ending the call with Little Apple, Xiang Kun started picking up the boxwood wood at home-he went out once last week. In addition to replenishing the rabbit, he also found old sources to buy a bunch of boxwood wood of different sizes.

At this time, these woods came in handy. Xiang Kun was going to carve them into some computer accessories, and then used the “interest” and “focus” emotion modes for “emotion injection”.

Before this week’s weekend, he was about to finish a blood drinking period, when these “emotional injection” items could just take effect.

Since this time, Xiang Kun has basically sensed the “Little Huang Woodcarving” of Little Apple once every two days, guided her to dream, and then “watched” in the dream.

At the beginning, after leaving the dream of Little Apple, he was able to use the special cognitive model of Little Apple to build a cognitive model. Only things that have appeared in Little Apple’s dream, contacted or heard by Apple.

However, after two blood drinking and two mutation evolutions, this cognitive mode has also become a faculty of Xiang Kun.

It was as if the software had only been installed before, but the hardware on this machine had conflicts, so that software could only be run in the server-Little Apple’s dream, and now after upgrading, his own “hardware” has begun to adapt.

Of course, compared to the cognitive model of Little Apple, it is still difficult for him to establish that particular cognitive model for any item. He can only have a very general feeling, mixed with his original cognition. Seems slightly confusing.

Xiang Kun’s “divine light flashed” when he was doing his “AI assistant plan”, in addition to the ideas emerging from his own mutation process and mutation model, there are those special cognitive models obtained in the dream of Little Apple .

He had a faint idea of ​​combining the two “things”, but he couldn’t find a good entry point. After all, whether it is a mutation of himself or the special cognitive model from Little Apple, he has not fully figured out what is going on. It belongs to the state of being indifferent.

So now he wants to guide Little Apple to learn about the computer, and then use her special cognitive method to build a cognitive model of the computer’s operating mode. Xiang Kun borrowed the dream of Little Apple to perceive the model and see if he could get any inspiration.

Compared to continually thinking about himself, Xiang Kun thinks this method may be more worth looking forward to.

Besides, he really likes the little apple child. He can teach her some computer knowledge. Maybe it will be useful in the future? After all, the information age.

On the weekend, I enjoyed a well-prepared meal at Xiang Kun’s house, played with “Golden Sparkle”, and was enthusiastically coached by Xiang Kun for several hours of computer-related knowledge, and gave away a bunch of his recorded ” “Audio Teaching Materials”, after getting three very special wood carving works, Little Apple and father Li Yang got on the bus home.

“The dishes made to Uncle are better than last time!” Little Apple said regretfully: “It is a pity that Uncle and us are not in the same city, otherwise we can eat rice from time to time, hehe.”

“Why didn’t you see it before, you are so greedy?” Li Yang looked at daughter said with a smile.

“Who doesn’t like to eat delicious food!” Said the little apple, carefully touching a wooden box in his hand, and said, “There are still these woodcarvings, and it feels very difficult to deal with, those capacitors are so small, the circuit It’s so thin, it can carve traces to Uncle. It’s really amazing. “

In the wooden box given to her by Xiang Kun, there were three wood carvings of different sizes. They were the motherboard, CPU, and memory. That big piece of wood.

When Xiao Apple was at Xiang Kun’s house before, Xiang Kun also deliberately dismantled a host that he had previously bought for the project. He guided Apple and touched the internal components of the computer such as the motherboard and CPU in detail. She introduced where she touched and what the function was, as opposed to what he had taught her on WeChat.

Looking at the wooden box in the hands of Little Apple, Li Yang also lamented: “Xiang Kun is indeed a good man. You can feel that he really wants to teach you computer knowledge.”