This is how black technology should be used

Chapter 75 It's Not a Question of Strong or Not

Artificial Intelligence or Artificial Intelligence?

this is a problem.

In the past 20 years, artificial intelligence technology has gradually been deeply integrated with computer technology and the Internet.

Thanks to the development of the four major catalysts of massively parallel computing, big data, deep learning algorithms, and human brain chips, as well as the reduction in computing costs, artificial intelligence technology has advanced by leaps and bounds.

It takes advantage of the development opportunities of computers and the Internet, aliases business intelligence, data analysis, informatization, automation, etc., and penetrates into every corner of social development.

On the one hand, the promotion of the Internet has created many application scenarios for artificial intelligence, reflecting its real value.

On the other hand, the upgrade of computer software and hardware has provided artificial intelligence with powerful computing power. Algorithms that could only be realized in theory before have been implemented, allowing artificial intelligence to create miracles in more and more events, and even surpass humans.

But when it comes to practical applications, the most criticized part of artificial intelligence is: artificial intelligence does not reflect intelligence.

Many people's perceptions of artificial intelligence are divided.

On the one hand, the media keeps reporting what kind of new achievements artificial intelligence has achieved, various big names abroad remind people to be vigilant about the development of artificial intelligence, artificial intelligence is included in my country's development plan and other big news.

On the other hand, there are always accidents in autopilot in the news, smart furniture at home behaves like a mentally retarded person, and information platforms are always stupidly pushing the same type of news...

These phenomena make people wonder, where is the intelligence of artificial intelligence?

For example, when the unmanned car accelerated and crashed into the overturned truck.

Because the white light on the roof of the truck makes the camera of the unmanned vehicle produce a blinding reaction, the AI ​​will not respond without any information, and may even give a wrong conclusion, mistaken for a smooth road, and an acceleration go up.

Moreover, the "reflection" is also inferred from the accident process after the traffic accident, rather than the conclusion directly drawn by AI.

AI will not tell you, "The reason why I bumped into you was because I was careless and didn't dodge."

There is also artificial intelligence in traffic, which can only accurately recognize traffic lights, but cannot recognize advertising posters on buses, and directly frames the stars on the posters, thinking that they are running red lights.

There is also a website's intelligent recognition picture that directly recognized black people as gorillas, was sued for racial discrimination, and almost triggered zero-dollar purchases.

The working principle of this image recognition technology is to convert patterns into digital codes, then find features from these digital codes, look up the "dictionary", find the corresponding explanation and then display it.

In fact, the computer does not know whether it is a black man or a gorilla, but the "dictionary" tells it that this feature has a high probability of corresponding to the word "gorilla".

Most of the algorithms are essentially playing games of probability. The only difference is that the information required for model training is different, and the calculation methods corresponding to the "gorilla" are different.

All well-known models that are widely used at present obtain certain probability distributions through matrix operation training data.

The probability distribution of complex models is usually high-dimensional, and various mathematical methods will be derived here, but the essential idea is still to describe the characteristics of the training data through the probability distribution.

With these, the same probability distribution can be used to describe the same kind of data, so as to realize the so-called "recognition" or "prediction".

In fact, it is not that the model really understands what a "gorilla" is like a human being, but in this way, the model can identify pictures that look like certain specific targets with a high probability.

Many people now divide artificial intelligence into strong artificial intelligence and weak artificial intelligence.

Among them, strong artificial intelligence is the "legendary" artificial intelligence, and many people even pursue to let computers have human mind and consciousness, and have independent choice behavior.

However, the research of strong artificial intelligence is more difficult, and there is no mature application on the market.

Narrow artificial intelligence is more like a tool to solve specific problems.

The characteristic of this type of problem is that it can summarize experience and form a solution through statistics, and this method of solving problems is called "machine learning".

And the more accurate the data, the better, and the simpler the better, once the data becomes blurred and the complexity increases, it will immediately change from weak artificial intelligence to artificial mental retardation.

For example, quantitative trading, face recognition, and AlphaGo are all machine learning models that are good at a single aspect.

When training the model, the technicians only taught Alpha Dog the skills to play Go, so it can only play Go.

If you throw a math problem to it, it will be confused immediately.

Just like the current members of the visiting group, they are also a little confused.

Most of the visiting group are technicians. Those who are not technicians have also carefully understood the information of practical technology. Some technicians' research fields are somewhat intersected with artificial intelligence, and they all know some knowledge about artificial intelligence.

They have never seen such a powerful artificial intelligence.

Smartness comes...it is not a question of being strong or not, but rather special. Anyway, it is not weak at all, and it has not been heard that it is mentally retarded.

Li Wei just said to prepare the car, but didn't say how many people they have, and then they prepared such accurate vehicles. How did the company's management AI determine the number of them in the corridor?

Why didn't you include the employees passing by?

But this time there are not enough people with related majors, and this is the topic of the next cooperation, so they didn't say anything.

But when the vehicle was moving, Li Wei gave a brief introduction.

"There are not many auxiliary management AI functions at the company level, but it is like the lubricant in the machine, which can fully lubricate the various joints of the huge machine and make the machine run more smoothly."

"For example, the cars we are riding in were originally ordinary park patrol cars. We modified them, but they are still conventional unmanned vehicles, even simpler than the unmanned vehicles in the world."

"But it doesn't matter, it's not capable enough, we can add some assistance, and then control all the company's cameras, so the unmanned vehicle can not only identify the route by itself, but also use the company's cameras to help it provide additional information."

"Look at the screen on the car," Li Wei pointed to the screen on the car and said, "The picture here is captured by the car itself, and the picture next to it is provided by you."

"There is a corner ahead, and many employees are walking towards it. For the vehicle, the employees at the corner appear suddenly, which is a test of the emergency handling ability of the vehicle AI. If it is not handled properly, a car accident will occur."

"But there is an oncoming assistance. The vehicle already knows that someone is coming, so it can make a correct response one step in advance to avoid car accidents."

"It is very helpful for the coordination and cooperation within the company..."

While talking, the electric car took them to the R&D department, and when they first entered the R&D department, they saw Li Weilai directing a "steel warrior" to walk.

"Don't move anything, just treat it as paralyzed, and let the machine take you away."

"But I left involuntarily! Consultant Li, I'm not paralyzed, but my legs just don't listen to the brain. It, he has his own thoughts."

"..."

　Thanks to Tian Cang Ling and Shu Huang for the reward. I had to write it myself~~Second update at 12 o'clock

　　
　
(End of this chapter)