Photo source @ Visual China
Text | Luo Chao
A few days ago, I saw an article "algorithm is chip, release is commercial, XX technology, which is different from traditional impression, sent a cloud AI core", and put forward a new concept: "algorithm is chip". After reading the article, it is stuck in the throat.
This article does not target anyone, but involves the company in question using a pseudonym.
What is an algorithm that is a chip?
What does "algorithm is a chip" mean? Let's take a look at the official definition of this company:
Moore's Law is near the end, and the increase in the number of transistors per unit area will eventually touch the physical limit, making it difficult to accommodate more computational power in a smaller space. However, the performance of the intelligent algorithm has increased by ten thousand times. Taking the face recognition algorithm developed according to the graph as an example, the accuracy of the algorithm has been improved by 1 million times in the past four years. In this context, artificial intelligence promotes the computing industry into the era of "algorithm is chip".
The person who proposed this concept thinks:
"The algorithm, that is, the chip is very essence, only can find the problem, look for the scene, use the pair of algorithms, and to customize the chip for this purpose, it is possible to achieve the ultimate price/ performance."
The third party interprets the concept as follows:
From the perspective of Moore's Law, the physical growth of computer architecture is approaching the limit, but the growth and optimization space of the algorithm is indeed infinite. On the one hand, the upgrade of the algorithm can improve the physical architecture and improve the physical performance of the chip; on the other hand, the upgrade of the algorithm can bring about a leap in computing power. In other words, the algorithm determines the room for growth in chip performance.
In the spirit of rigorous verification, Luo Chao Channel deliberately searched Google with keywords such as “algorithm as a chip” and found that this concept does not exist in foreign countries. It seems that it should be the first of “XX Technology”.
Is there any reason? Of course there is. After Moore's Law no longer works, chip performance and computing power increase, and more and more dependent on algorithms. Whether it is Apple A12 bionic AI chip or Huawei smart chip Kirin 970, the core idea is to put the AI part of the calculation on the chip. In addition to Apple and Huawei, Baidu, Ali and other companies have launched chips with built-in AI algorithms, and the application scenarios are dominated by the cloud.
In theory, this reflects the trend of AI computing: cloud AI terminal AI (and edge AI) is combined to adapt to the requirements of AI computing in different scenarios, but the concept of "algorithm is chip" is proposed, which makes people feel that although they do not understand, it is very good.
"algorithm is chip" is a bit of a piece of crap.
"AA is BB", which is used first in the software industry. The best known is Software as a Service (SaaS). The more direct point is software service. The cloud computing market extends PaaS. Different modes, such as (platform as a service) and IaaS (infrastructure as a service), have come to the Internet industry, and the concept of content as a service has emerged, which is easy to understand.
The essence of "AA is BB" is to carry the value of BB in the form of AA, for example, to carry the computing service behind it in the form of software, instead of saying that both things must be related, we should say "AA is BB".
Some people like to use "XX is XX" to create concepts. This trend has occurred in the blockchain industry in 2018. In the first half of 2018, the founder of Fcoin proposed the concept of “transaction is mining”, and soon there appeared various “XX mining”. For example, the APP reading the novel said, “Reading is mining”, reading More coins are sent more; shared bicycles say "riding a bicycle is mining", riding a bicycle to send coins; McDonald's launched "eat hamburger mine", eat a Maccoin when eating a giant. . .
The problem of "algorithm is chip" is that the algorithm is the algorithm, the chip is the chip, one soft and one hard, plays its own role in the IT industry, the chip is responsible for the calculation of the algorithm, although the chip has built-in algorithm, but can not say that the algorithm is the chip. The relationship between the two is different from the relationship between software and services, and the two can not be confused.
We can't say that the algorithm is more and more important for taking pictures of mobile phones. We say "algorithm is the camera"; we can't say "algorithm is software" because the algorithm is more and more important to the software; we can't because the AI is more and more mobile phones. Important, it is said that AI is a mobile phone. "Algorithm is a chip", just like "egg is a cucumber", it makes people feel a little bit.
Why are companies passionate about making concepts?
Why do enterprises always throw out all kinds of new concepts?
The most important role is to master the right to speak.
The second role is to attract attention from all walks of life.
Including users, customers, capital and media, it is more straightforward to communicate better with the public and better with public relations. Creating a new concept can attract attention, capture labels, strengthen public awareness, and make difficult business stories easier and clearer.
The core purpose of XX Technology's "Algorithm is Chip" is to emphasize that its algorithm makes the chip powerful and attract potential customers. Last year, many companies embraced the blockchain to attract capital attention; the Internet industry now values young people, and the "Z generation" is favored by many companies. .
The new concept is the future. The foreseeable enterprises are far-sighted and enhance their image. I remember that when the concept of "Internet +" was popular, many companies came out to claim that they should first see such a trend, and the essence is To enhance the corporate brand image.
The fourth effect is pure hype.
At first, Letv of Jia Yueting had a concept of "ecological inversion," ecology, business matrix, and cross-border concepts, which were talked about a lot, but Letv put forward a new concept, that is, chemical reaction, which was different at once. This is even more evident in many A-share companies, such as companies that directly change their name when P2P is hot.
Many concepts are not necessarily original by enterprises, many are created by media, experts and scholars, and have been adopted by enterprises. No matter what the way, the enterprise makes the concept and uses the concept, all needs to have the correct posture.
The correct posture of the concept of Enterprise Construction
First list a few concepts that I think are good: new retail, social e-commerce. Where is it?
1. Easy to understand and remember.
New retail only uses a "new" word to define a new form of business. At that time, Ma Yun also put forward new finance, new manufacturing, new technology, and so on. Now there are "new" in many fields, such as "new marketing." I have to say, Alibaba is a master at building concepts, whether it is "eWTP", "digital economy" or "commercial operating system". Each is based on existing concepts (WTO, economy, operating system) and then evolves to make it easy to understand and remember. Many people say that Alibaba public relations is a public relations group, which is also the embodiment of its strength.
Good concepts are vivid, easy to understand and remember, and the public will know or guess the meaning behind them. A bad example is a concept similar to VLog. To understand an article, you don't necessarily understand it.
2, the old is not as good as new.
After some companies have proposed Internet +, some companies have thrown out the concept of "+Internet". Although it does explain the difference, it is difficult for the public to understand the nuances of "+" before and after, and "industry Internet" is big. There is a reinvention, and the "five new" thrown by Ali is essentially the landing of the industry Internet in some key industries. After the "micro-business", there is a concept of "social e-commerce". The essence is one thing, but it seems to be more pressing. It is a new concept of success. After "social e-commerce", it will do more "social grouping". Later, there was a “community group purchase”, which was a bit difficult to understand. After all, companies must dare to introduce new ideas with new concepts and start a new one.
3, can be self-explanatory.
The concept of the concept itself should be logically supported. It can be said that the problem of "algorithm is the chip" is that it does not make sense. Including some shared economies before, it does not have the "sharing" trait itself, and it is necessary to pull on the concept of sharing economy. Hard to accept.
4. Successful business landing.
The most important thing is to have a business landing, otherwise the concept no matter how good it makes no sense. Blockchain should be one of the most conceptual industries, from digital money to ICO to public chain to alliance chain to ledger. Although I am a technical learner, many endless concepts of block chains are well beyond my understanding. In 2019, the industry clearly felt that blockchains were no longer so hot.
The Internet industry is experiencing new concepts every day, but what really drives the Internet forward is never the concept itself, but the concept of the concept in the business scene.
A good concept is important, but it's just an icing on the cake.
More exciting content, focus on Titanium Media WeChat (ID: taimeiti), or download Titanium Media App