ChatGPT breaks the language barrier between people and computers

The development process of science and technology is to continuously break the boundaries between humans and nature.

The emergence of chatGPT breaks the language barrier between people and computers.

Computers are based on digital circuits, while humans are based on biochemistry. The underlying mechanisms between the two are different.

Although both computers and human brains can be used for information processing, it is not easy to break the language barrier between them.

1, C language only partially breaks the language barrier between people and computers

C language appeared in 1970. It is a programming language similar to English, but its syntax is Human language is much stricter and does not have any vague semantics.

Fuzzy semantics is one of the basic characteristics of human language.

Before chatGPT, any "language" used to communicate with computers had to be strictly precise.

Whether it is C/C or java/python/js/php, they can be converted into binary codes and correspond to the status of the digital circuit one-to-one.

In short, programmers used to accommodate computers, but computers cannot accommodate humans.

Natural language processing (NLP) has been developed for so many years, and it is nothing more than matching some keywords. There was no breakthrough before chatGPT.

2. Language is probably the final boundary between man and nature.

This is not the first time that the boundary between man and nature has been broken. In fact, it has been broken many times before.

For example, people in the 18th century believed that urea (H2N-CO-NH2) could only be produced by biological metabolism and could not be synthesized by chemical means, so such substances were called organic matter.

But in the 19th century, chemists used carbon dioxide and ammonia to synthesize urea. This was the first time humans synthesized organic matter.

The synthesis of urea breaks the "organic boundary" between humans and nature, that is, organisms are not special in nature, and substances that can be synthesized by organisms can also be synthesized through chemical means.

Of course, urea is just a very common chemical fertilizer now, but when it was first chemically synthesized, it was still very shocking, because the "superiority" of humans as living creatures was challenged by chemists. Now that chemical methods can synthesize organic matter, what else can humans do that chemistry cannot?

Logical ability. People have logical thinking ability and can make if else judgments. However, no mechanical equipment could have the ability to make logical judgments in the 19th century.

Then, the electronic triode appeared.

In 1946, the ENAIC computer was invented, which could run logical instructions written in machine code much faster than humans.

After the advent of the transistor in 1947, Moore's Law was initiated, and the performance of computers began to iterate rapidly!

Then, people have to find something that humans can do but computers can’t do to re-establish their “advantage”.

Before 1970, what could humans do that computers couldn’t?

Computers can only understand machine code and assembly language, but cannot understand high-level languages ​​(similar to human language).

Then a bunch of high-level languages ​​appeared, such as FORTURN, pascal, C, C, java, perl, python, js.

Computers can recognize high-level language codes, but computers cannot see things (CV), nor can they understand real human language (NLP). This was probably the situation before 1980.

3, CV and NLP have always been the traditional advantages of human beings

There is such a joke: the product manager asked the programmer to write a code to identify colors, and the programmer replied, "Do you think App It's a dog, it will recognize colors after training."

This joke shows that the early CV model was very ineffective, not to mention compared with people, and it was incomparable with animals.

With the development of deep learning after 2016, great progress has been made in classifying complex objects through CV models. The accuracy of face recognition has been greatly improved, but the recognition of language has always been There is no decent breakthrough.

Before chatGPT appeared, no one wanted to chat with a computer because the computer would not answer after a few words.

But now chatGPT has appeared, it can answer most common questions and even give C codes for some algorithm questions.

Human language advantage has been challenged again by technological progress, and silicon-based intelligence has taken another big step forward in the direction of "silicon-based biology".

Once an object has "real intelligence", no "person" will be willing to just be an assistant.

Les-Frostwhisper in "World of Warcraft" became a lich. Is he alive or dead?

No wonder Musk and others are worried about continuing to train more powerful AI models.

The above is the detailed content of ChatGPT breaks the language barrier between people and computers. For more information, please follow other related articles on the PHP Chinese website!