Ai-face object language (1)

I suddenly felt that it was really tiring to write a program. If there was an intelligent software that could help me write the program, it would be nice.

So I started to think about it. For example, I could come up with a program like this.

All the software running on the computer (except for digital operations and chatbots) is a summary of regular tables, and they have been repeatedly searching for matching and result output. Of course, if you know the smart software, please let me know.

The key question comes: What is the core of "intelligence" (AI)? I have not been defined by more authoritative organizations. I think it should be "learning ability" for an afternoon ".

Imagine a baby born with nothing, no thinking, no language, everything is blank, just curious. (Aside, modern science can only prove that babies do not understand anything, maybe not in a few years), why do they speak, sing, count ...... What about it?

The answer is of course the learning ability.

What is "learning ability"? Similarly, it is not authoritative. What you think should be in a language close to the computer:

1. Memory

2. Load the historical memory and select the conditional reproduction output.

Memory capability. This is relatively simple. We can use ctlr + S to save files to the disk when using word. This is the memory capability. Of course, this can only be understood as a "rote memorization" Method for humans. But it does not matter. With this basic capability, we can combine more with the second one, so we will not go into depth.

"Conditional selection" is actually "logical Capability ". What is this "logical Capability"? It also generates a system combination capability through "learning ability. Currently, computers do not have this capability. We need to implement it.

This capability, starting from the perspective of imitating human brain functions, seems more difficult, because now humans have a limited understanding of the functions of the human brain, or "humans" are a bit too much, "I" is very appropriate.

It does not matter how this logic capability is implemented.

First, let us assume that the program that can "Learn" has been implemented. That is to say, the logic capability is also available.

The following two conditions are met: 1 and 2. We have a "baby" brain program. Next we should teach this "child". In sci-fi movies, robots are learning fast. Watching movies, reading books, and reading ten lines of books is not an exaggeration, this is just the "Memory" capability of normal computers.

We should also teach our "baby brain program" in this way. Let's call it Nao later, so that it can quickly learn and exert amazing capabilities.

Wait a moment, the program is very simple and cannot recognize images, because the children do not understand movies or TVs at the beginning.

Speaking of features: it's easy to teach him to speak and pronounce from the very beginning. Store the Chinese Character Font in the computer, and install a voice distribution software for the Nao.

Listening function: There are a lot of language sheet input software. If they all have APIs, OK.

Write function: Isn't text output to the textbox box? Print at most. OK.

For the time being, the functions such as eating, drinking, and playing are not considered.

Our Nao is very intelligent. It is easy for him to learn Chinese characters. But he does not really understand every word.

So what is "understanding?

To be continued...