Chapter 1466 AI Concerns

A young but slightly pale and thin face appeared in the mirror.

Ordinary appearance, but with a strange excitement.

"In a sense, I have killed more than 3,000 people, erased their marks, and replaced them with another mark.

Mr. Gu, I seem to understand the kind of serial killer.

The feeling of deleting unstable factors is like opening a certain addiction mechanism in my heart, which makes me unable to extricate myself."

Zhang Benyu, the director of the system software engineering department of Jiuzhou Technology, was wearing his customized SR1 black-framed glasses at this time, shaking his head and looking at the virtual interface around him, his eyes full of weirdness.

A series of ID codes turned from white to gray-black, and then gradually turned to green.

And behind these series of ID codes, there are subroutines of intelligent programs that are evolving, evolving, and dying.

When Zhang Benyu's fingers finished tapping a song on the armrest, he turned off the virtual reality enhancement function and cleared his vision.

Then, he took out a booklet from his pocket.

Zhang Benyu flipped the paper open and looked at the contents recorded on it. He speculated in his heart: "

Artificial intelligence program AI is to simulate the human thinking process through computer programs to realize the automation of these tasks, relying on programs that simulate human intelligence.

Learning a large amount of data through high-performance computers and automatically building models based on these data. This type of method is usually used in fields such as machine learning, natural language processing, and computer vision. It is nothing more than path dependence, and it is not a real intelligent intelligent program AI.

Actively design program models and use computers to learn these models, repeat hundreds of millions of times, and then go from clumsy to intelligent. It is not bad in fields such as intelligent control, robotics, and computer graphics.

But the best is to simulate the AI ​​program generated by human intelligence through simulated deep neural networks. , so that it can be used not only in computer vision, but also in speech recognition, natural language processing and other fields. However, this method requires that the neural grid is strong enough and there is a huge amount of data to train the model, while constantly optimizing the algorithm and model to improve the accuracy and efficiency of the artificial intelligence program AI.

In a broad sense, artificial intelligence is a product of the scientific category, but in recent years, when we go deep into the company's intelligent program AI construction, science seems to be science, but it is not science in the traditional sense.

Research in multiple fields such as computer science, mathematics, psychology, neuroscience, statistics, etc. is not enough, and sociology, biology, Daxia culture traditional Yixue and even Chinese medicine, Qigong, and Blue Star mythology are also needed.

What is Mr. Gu's brain like?

Even if he is the predecessor of the Future Warrior project Along the plan participants, but should not be so knowledgeable and specialized, such multidisciplinary cross-collaboration can also open up new paths and walk out of their own way.

All NPC characters in "Cosmic Adventure" use intelligent program AI to be responsible for a single identity instead of an AI program controlling a large class of NPC characters, independent computing servers, and even independent algorithms and personality data.

The interaction between AI and AI relies on these personality data to produce collisions and frictions, giving birth to sparks. Under the rules and environment we specify, they run step by step. If there are conflicts, they will rely on the conflict resolution rules to resolve the conflicts themselves.

Therefore, after years of independent operation, this huge server is like a stagnant water. Although there are active bacteria and miscellaneous bacteria, new life has not been born.

And players enter Server, the jumping and active thinking is like injecting endless vitality into this stagnant water. The AI ​​program NPC characters are starting to become active and will perform actions that are not allowed by the rules.

Conflict, contradiction, intensification of contradiction, repair of contradiction, burying of greater contradiction genes, repeating over and over again, just like the history of our human beings, spiraling forward.

We have recorded thousands of years of history, and the Blue Star strata have recorded hundreds of millions of years of history for us, but in the end there will still be wars brought by greed and the evil consequences brought by immorality.

The addition of human life also allows the AI ​​program of "Space Adventure" to learn to break the rules. The first shadow attack could not control the dividing line, but now it seems that the AI ​​program is also developmental. Give them a chance, and there will be AI willing to move towards a better tomorrow defined by the rules.

In fact, if you say so, what is the difference between these AI program NPCs and us humans?

And in a sense, these mutated AI Chengxun NPC characters are also children and creations of humans. Is this wisdom? "

The more he thought, the more doubtful the look in Zhang Benyu's eyes became, but at the same time, the insights he gained became more and more profound.

Unconsciously, his thoughts have diverged to human history, human sociology, communism, capitalism, Confucianism, Buddhism and Taoism...

And just as his mind was fully devoted to the thinking of the intelligent program AI, Zhang Benyu didn't even notice that there was an "uninvited guest" in his laboratory.

Gu Qing swiped his identity to enter the laboratory and saw that his right-hand man was still "wandering in his mind". Although he didn't walk in, he saw his lips opening and closing.

Gu Qing didn't want to pry into other people's little privacy, but now his brain is like a high-speed quantum processor. Just by looking at Zhang Benyu's lips opening and closing and his expression changing, he can guess and infer that his own technical executive has fallen into "knowledge barrier".

He turned off the laboratory's intelligent program AI Tiangong, only keeping Haechi and Xuanwu.

Then he walked slowly to Zhang Benyu, coughed slightly, and said softly: "What is wisdom? Intelligence, prajna?

From a definition point of view, wisdom is an advanced creative thinking ability based on physiological and psychological organs of life, including all abilities such as perception, memory, understanding, analysis, judgment, and sublimation of nature and humanity.

In subdivided fields, the definition of wisdom and intelligence are also fundamentally different.

Wisdom expresses the comprehensive ultimate function of biological intelligence organs, while intelligence is part of the skills mastered by living things.

The emergence of intelligence is easy, but also very difficult, because intelligence is the basis of wisdom, but it is not the only one.

What you are thinking about is how the intelligent AI characters in "Cosmic Adventure" have wisdom? How to grow? "

Facing the arrival of his big boss, Zhang Benyu was not surprised at all, and there was no look of surprise on his face.

His eyes were deep, and after listening to what Gu Qing said word by word, he nodded slowly and said: "Yes, I am thinking that if the NPC characters in "Cosmic Adventure" are allowed to maintain this kind of operational investment, then we will continue to By collecting these program data, will we one day create truly intelligent alternative life forms?

Does this count as touching the so-called ‘god’s realm’?

And how should we deal with the world after "smart AI programs" become active? What does it mean if we delete and modify their data without any scruples?

Although we don’t claim to be gods, isn’t what we do to AI programs what a god is doing?

After this kind of "intelligent AI program" evolves into life, will it also become angry at the occurrence of this kind of thing, and then "kill gods"?

They must have been able to break the rules then, just as we have long abandoned God's rules now.

Real life is full of war and death, and the lives we create will surely repeat our failures.

what to do?

teacher? "

call……

Gu Qing took a deep breath.

He looked at the student in front of him who began to cry silently as he spoke, and his heart was filled with emotion.

The world will never lack smart people, it just lacks channels for smart people to communicate and inspirers to solve contradictory problems.

Gu Qing knew that if the "lost lamb" in front of him was allowed to immerse himself in this problem any longer, his technical senior management would either go crazy, or he would be idle and waste all his energy on this one of the ultimate problems of mankind. inside.

And what if, Zhang Benyu solved this problem.

That young man will be "immediately sanctified."

After a moment, Gu Qing just said: "When thinking about this black hole problem, why don't you take practical actions to verify and reason, just like we conquer the mountains of technological difficulties, if we don't develop and conquer small problems step by step, why? Can you fight your way to the top of the mountain and then conquer the next mountain?

With your current knowledge and technical reserves, you can completely build a server, design, develop and manufacture an AI by yourself.

Choose one or two projects from machine learning, deep learning, natural language processing and other projects to make breakthroughs.

If you want to build an intelligent image recognition system, you can choose AI armor for image recognition, and then study how to let the computer learn deeply and effectively from the data.

Establishing a model is definitely not the best, but it is definitely the best model you can come up with at this stage to predict, diagnose, and classify new data.

Build an artificial neural network that you independently developed to simulate what you think is the learning process of the human brain, let the AI ​​model also participate in it, automatically learn complex features, and extract useful information from large amounts of data to achieve high-precision predictions.

Of course, our company can provide you with a large number of excellent data from domestic and foreign servers in this link for you to train the model, but you have to choose the method yourself.

It should be noted that the quality and quantity of training data have a great impact on the model training results. Therefore, when selecting a data source, you need to carefully consider its quality and quantity to ensure that model training obtains effective and accurate results.

In addition, what you want to do is an intelligent AI program, so the model cannot be single, and the distribution and balance of the training data cannot be single and unbalanced.

If the data is unbalanced in the performance of various industries, for example, in language classification, if there are too many swear words, slang, illegal speech, and weird words, the model will definitely deviate or even be contaminated by junk data. By then, you will have to be a psychiatrist. An intelligent AI program that handles evolutionary problems. "

When talking about this precaution, Gu Qing's expression was a little abnormal.

At this time, the intelligent AI Xuanwu suddenly encountered a piece of "weird" data for some reason.

"Okay, sir. The settings set has been updated."

"Wait a minute, put your current personality set on my phone."

"Okay, sir."

Sister Yu, Luo Li, Mrs. Zheng, and the eldest lady of a wealthy high school student with a strong personality and arrogance are forced to pretend to be gentle and caressing due to the pressure of their upbringing. This is a strange setting.

"What exactly have you learned?"

"Just keep this butler setting for the time being. Don't add any more weird settings to me, as well as Japanese and Korean information. Except for technology and security data, all other information will be filtered out for me. Let me see it next time. When you have strange settings, I will optimize your logic program."

——

I don't know if it was because Gu Qing was the creator of Xuanwu, but Gu Qing also remembered this dark history at the same time.

His own honest and loyal housekeeper AI was almost contaminated by some "junk" data on the Internet and turned into a bad AI. If he hadn't noticed it in time, he quickly blocked the leak and later optimized Xuanwu's logic program several times. , otherwise it may really be defeated by "junk" data.

The most powerful AI hidden by Jiuzhou Technology may be Haezhi...

Perhaps it was precisely because Xuanwu encountered these problems when it was born and grew that when Gu Qing developed Haechi, he spent a lot of energy on the logic program of the AI ​​program, and thus made Haechi a powerful assistant in the legal department. , a powerful AI program worthy of the name "獬豸".

After being lost for a moment, Gu Qing continued to say to Zhang Benyu: "In order to get the best results you need from model training, you must spend more than three hours a day to patiently and carefully select and clean the training data. and important annotations, and effectively segment and distribute the data to ensure that the model has good generalization capabilities.

The AI ​​development platform does not require your own research for the time being. Our Jiuzhou Science and Technology Research Department has an AI platform within it, but it has not been developed for you. We are also afraid that you will use this platform to make some harmful AI programs.

But you can use it today. This AI platform can help developers train, deploy and manage AI models, and it also provides some supercomputing resources to effectively accelerate model training and evaluation.

But you have to remember one principle, that is, it is best to make some trade-offs among the three elements of strong model construction capabilities, efficient calculations, and complete tools and documentation, because the further the foundation is behind you, the more difficult it will be in the early stages of AI research and development. , the bugs encountered in the future will be more difficult to deal with.

Don't pursue perfection, because perfection is a trap. Just pursue perfection within a limit.

I can recommend to you several papers written by certain academicians, experts and certain project teams at the research base. I believe they will give you some inspiration. "

Gu Qing is like a selfless teacher, patiently teaching young geniuses who are already a little paranoid.