Ask not what your country can do for you, ask what you can do for your country.

(Do not think about what the nation can do for you; think about what you can do for the nation.)

John F. Kennedy

Introduction

Starting with deep learning, artificial intelligence is suddenly becoming exciting. I do not think most companies, except those with enormous capital, are yet at the stage where they should move as companies, but I am personally paying attention to it and playing with it. I decided to write down what I am thinking now, as a record of my current understanding.

Masumi Kawasaki Blog I used deep learning to make a photo of my daughter look like Van Gogh masumikawasaki.blogspot.jp

Is artificial intelligence real?

ついに歴史的な日が…。囲碁のプロ棋士、グーグルの人工知能に破れる _歴史の1ページに、新たな項目が加わりました。 グーグルが AIの研究に注力していることは知られていますが、彼らが開発した 囲碁ソフト、その名も「 AlphaGo」がついに プロ棋士を打ち負かす という、衝撃のニュースが飛び込んできました…_www.gizmodo.jp

I play Go a little, and the speed of this development is beyond what I expected. This artificial intelligence technology is made using a technology called deep learning, but I think we need to make our understanding of deep learning a little clearer.

That is because when people say artificial intelligence, it creates the idea of “intelligence the same as humans = something that replaces humans,” and the discussion moves in a different direction. Many discussions about artificial intelligence do not have a fixed point of argument, and they resemble religious debates about whether God exists.

Also, as with recommendation engines, we have repeatedly called new technologies artificial intelligence and then given them different names once they became widespread. Therefore, it is important to understand the basic parts of the technology called deep learning.

So what is deep learning? It is often said to be “modeled on the human brain” and so on. That is correct as a way of thinking about the implementation, but it is not a phrase that expresses the benefit users receive. So, without fear of being misunderstood, I think it is a technology for recognizing information and making judgments.

Through it, we can use massive amounts of data that humans could not process (recognition), and produce answers that far exceed human ability (judgment). Is that not the essence of deep learning?

Aside 1: The speed of AI development

As an aside, I am also concerned that many people talk about the speed of this technology’s development together with its technical superiority. This speed of development is largely due to the development of the internet and has nothing to do with how amazing deep learning itself is. A single idea is having a large impact in a society where information circulates easily. This also relates to the singularity, but I will leave that aside this time.

However, it is also true that because deep learning, a “technology for recognizing information and making judgments,” is used on top of the internet, a system where information is highly open and exchanged, its effects are immeasurable.

From an information society to an intelligence society

Another characteristic of deep learning is that even the thought process for judging information can be left to the machine. Therefore, by importing a large amount of data learned in the past with a rough direction, it can derive an answer. There is no need to define the thought process. It is close to how humans make intuitive judgments.

(On the other hand, this hides the thought process, which is also dangerous.)

And because it is a technology rather than a person, it is reproducible, which makes it powerful. What happens as a result is a state where you have multiple personal brains, multiple personal consultants, and multiple personal concierges.

In other words, the world is moving from an era of using information through the internet to an era of using intelligence through recognition technology.

Aside 2: Reference books, or manga, for facing artificial intelligence

Just as Ghost in the Shell foresaw a coming future through the themes of self and information, I recommend reading Prophecy Pippi, although it is unfinished, when thinking about the future form of artificial intelligence and humans.

攻殻機動隊（１） (ヤングマガジンコミックス) _西暦2029年。通信ネットワークに覆われ、膨大な情報が世界を駆け巡っている超高度情報化社会。しかし国家や民族、そして犯罪は依然として存在していた。より複雑化していく犯罪に対抗すべく結成された特殊部隊…公安9課に所属するその組織…_www.amazon.co.jp

預言者ピッピ１ (CUE COMICS) _「つまりぼくら、人類の未来をすべて計算しつくしちゃったんだ」究極の頭脳が紡ぎ出す預言は福音か、それとも…？鬼才が描く長編ストーリーが、満を持して登場！ピッピは地震を予知し、災害から人類を救う為に開発されたロボット。親友・タミオ…_www.amazon.co.jp

Will artificial intelligence threaten us?

NO. Of course, there will be concerns such as people abusing it or transferring it to the weapons industry. But the person is bad; the technology itself has no good or evil.

There is no doubt that this technology has benefits for humanity. We need to discuss how to use it from now on. Its use will likely progress first in industries that are easy to informatize, such as finance and journalism.

Will artificial intelligence take our jobs?

YES. Many of our jobs can be patterned, and jobs like that can be replaced by artificial intelligence. This had already begun before deep learning. For example, search engines probably took away many consultants’ jobs. On the other hand, there will be work creating artificial intelligence. We cannot stop technological evolution.

I was once asked, “When artificial intelligence develops, what work will humans be able to do?” At that time, I rephrased the question: “When artificial intelligence develops and we no longer need to work, what can we do using artificial intelligence?” We should not oppose technology; we should aim in the direction of using it.

By the way, I think the answer is “entertainment” and “education.”