After the layoffs, I became a tech godfather

Chapter 109 Will Xiaosha become conscious?
Latest Beta version.

In terms of ability alone, there has not been any significant improvement.

It is mainly the optimization of some comprehensive capabilities and there are some improvements in series connection.

After these three sentences, both Shen Wei and He Gang didn't know how to respond!

The ability has not been improved yet. With Xiaosha's current ability, is there any room for improvement? No, Hao Cheng didn't mean that there was no improvement, but that there was no particularly big improvement!

What Hao Cheng meant by "extraordinarily large" is completely different from what others mean by "extraordinarily large".

A word suddenly popped up in He Gang's mind: "humanoid".

I carefully recalled the scene of today's meeting. Suppose the person present was not Xiaosha, but a person with limited mobility who could not move temporarily, could this person do better than Xiaosha?
Hao Cheng said to start, and the curtains were closed and the lights were dimmed. The approval process came in, the video was paused, and the split screen was turned on. He accurately understood what Hao Cheng meant, marked the time in the approval process, and sent a message to Cao Song on WeChat. He called to order food and asked Xiaoyi to drive...

That's probably all a person can do, right?

When he thought of this, He Gang suddenly got a chill, and could not hold back the doubts in his heart, so he asked, "Mr. Hao, I suddenly have a question and I really want to hear your opinion."

"Hmm?" Xiaoyi was driving, Hao Cheng tilted his head: "What's the problem?"

"This has nothing to do with our cooperation or any business behavior. I just suddenly felt something." He Gang explained, and then asked: "What do you think AI can achieve in the future? Is it possible to awaken its own consciousness?"

This question is often asked by laymen because they do not understand the training of AI or how AI's seemingly magical abilities are "calculated". So they fantasize about AI awakening and ruling over humans.

Real industry insiders know that this thing is impossible to appear on the basis of existing artificial intelligence. The essence of artificial intelligence is actually a comprehensive extension of mathematics, logic, and computational science.

Although it is very complicated, no matter how complicated it is, it is nothing more than the word "algorithm".

But Xiaosha's performance today made He Gang hesitate and asked this question.

"Impossible!" Hao Cheng said with certainty.

"Why?" He Gang and Shen Wei asked almost in unison.

"The definition of consciousness determines it," Hao Cheng said. "The subject of consciousness is human. Even if a machine produces something similar, it is not called consciousness unless the definition is modified."

Hao Cheng deliberately made a joke and continued, "If you want to ask whether machines can produce something similar to human consciousness, if we define it from a biological perspective - in the adaptive activities of organisms to survive, the subjectivity produced by the developed central nervous system can be called consciousness.

"Things like this kind of consciousness can be easily produced by AI. I think Xiaosha is not far away."

Hao Cheng said this. If the first version of Xiaosha was not good enough, then after experiencing the upgrade of the grasp of "degree" last night, Xiaosha has obviously undergone a transformation.

The determination of right and wrong has gradually developed something other than rules and logic, but Hao Cheng currently has no idea what this thing is.

If you say that this is "subjective", then it is indeed not, but if you say that it is "quasi-subjective" and a reflection of "human subjectivity", Hao Cheng thinks there is no problem.

Based on this, one day, Xiaosha will definitely be able to produce something like "consciousness".

"For humans, consciousness is the reflection of the external material world by the human brain. It is the sum of various psychological processes such as feeling and thinking, and expresses human emotions, pursuits and will."

Hao Cheng continued: "But for AI, these things do not exist. But it will produce something. What is this thing?
“I think it’s a similar consciousness to that of humans.”

Hao Cheng was not stingy in saying more about this kind of general concept, and he even asked a question in return: "Converging with people? Who are you converging with specifically?"

"Convergence with the trainer!" Hao Cheng said a lot. Although he had never heard of this concept before, both of them were not outsiders. One was a telecommunications major and the other was a circuit system designer. They also had some understanding of AI in recent years. It was easy for them to come up with these general ideas.

"Yes." Hao Cheng said: "Whether it is traditional AI or Xiaosha, how it will eventually grow depends on what data and knowledge you feed it.

"If it does eventually develop a consciousness that is similar to that of a human, then this thing must also be similar to that of the trainer."

Hao Cheng also thought about this, because the training of "micro-consciousness" ultimately comes down to two words, namely "mapping."

This is the most critical core.

This is something that must be mentioned and reviewed in every late-night class. So far, everything Hao Cheng has come into contact with, including what he got after upgrading the "adder" he designed that day through energy points, has reflected the concept of "mapping" the most profoundly.

So, if AI develops a similar consciousness to that of humans, it must be a more complex mapping.

If human consciousness is the reflection of the objective world by the human brain, then the [similar consciousness] generated by AI is the reflection of the [AI network] on the [world input by the trainer].

Then, it is obvious who it is converging towards.

This reflection, when expressed in terms of the operating method, is still the [Super Code System] or [Mapping].

"Another more crucial point is that I think there is no need to worry about whether AI will develop this so-called [similar consciousness]."

Hao Cheng went on to say: "Even if it does, there is no problem. The most important thing is desire. As long as the machine does not generate desire, everything will be fine.

"As for the desire, I don't think it will happen, at least Xiaosha won't."

The reason why Hao Cheng said this is because the most critical part of the training method of [Micro Consciousness Body] is [Super Code System] and [Mapping].

How do human desires arise?
People get hungry, so they have a desire to eat. Throughout history, most uprisings were caused by famine and lack of food.

There are even many jokes that go something like this: If I travel back to ancient times and only feed the soldiers pickled mustard tubers and white rice every day, will they rebel?
If you want to rebel, they will put a yellow dress on you!

But does AI have this concept? Could it be that the voltage is low and AI says I am hungry? It is just that the calculation is slower.

According to the concepts of the "Super Code System" and "Mapping" in the late night class, there is none! Not only hunger and cold, but also emotions, sex, will, feelings, etc. are all gone.

It has passed the initial stage, like the little sand before yesterday. To grow further, it does need to simulate human will and emotions, and this simulation relies on the "mapping" of facts and logic. The complex "mapping" interweaves the concept of "similar".

Classes are different after all.

Therefore, it is impossible for AI trained by the "micro-consciousness" system to have desires.

This is the conclusion drawn by Hao Cheng.

As long as you don't have any desires, there is no need to worry about anything.

……

After talking so much with He Gang and Shen Wei, Hao Cheng himself suddenly had a deeper understanding of the [Super Code System], [Mapping], and even [Training of Micro-Consciousness].

When they arrived at the restaurant, Hao Cheng silently took out his mobile phone to record these things.

"It has to be said that mobile phones are getting bigger and bigger, and it is becoming more and more difficult to operate with one hand."

Due to the width of the Mate70 Pro+, even though Hao Cheng has big hands, it is still very difficult to operate it with one hand, especially when typing, as he can't even reach the left row.

The input method is fine, there is also a one-handed mode that can reduce the input method area, but if there is a button or icon on the other side that needs to be clicked, there is no way.

"If, when my finger goes over there, the thing I want to click on will come over by itself," Hao Cheng said while gesturing, holding the phone in his right hand with his thumb stretched out as if he was trying to reach the icon in the upper left corner:

"If at this time, the entire screen moves to the lower right, in the opposite direction of my finger, hey, when it reaches the middle, it overlaps, I click, snap, isn't it great?" Hao Cheng looked at He Gang and said his own "great".

"Wouldn't it be better if it could recognize what you want to click and help you click it directly?" He Gang asked this question.

"It can't be recognized!" Hao Cheng shook his head: "If your gesture sensor can directly recognize the action, it can point to where I focus my eyes, but it probably doesn't have such precision now.

"Also, it's hard to control your eyes, and it's tiring to move your eyes around. But I think it's not difficult to identify a rough area. You don't even need to use a rough estimate. Just divide the screen into three equal parts, and half of the left and right parts, and there are six areas in total.

"When my hand wants to operate, my eyes will definitely change accordingly, and then the area to be operated will be temporarily moved towards the finger, and then restored after the operation. This compromise should not be difficult to achieve."

Hao Cheng said it was not difficult to achieve, but He Gang knew his own business and he knew very well what the attitude sensor could do.

When designing the grabbing function before, that hand had to make deliberate movements.

Later, I used Xiaosha's AI capabilities to do the "shaking" action of shaking the files, and the amplitude of the action was also very exaggerated.

Now, recognize a finger and an eye contact?

Even if it is just to determine an area that is one-sixth the size of the screen, it is probably possible to do it, but the accuracy cannot be guaranteed. If the accuracy cannot be guaranteed, then it will be of little significance.

(End of this chapter)