The BOOM in AI, and the desire for it, has kept the titans enthusiastic. For example, Oxford University Press recently signed a partnership with Sogou to deliver Oxford Dictionary-related content. Baidu and Ctrip launched WIFI translator; Google Translate returned to The Chinese market last year.

At present, the leading translation companies are Baidu, IFLYtek, Netease Youdao, Sogou, etc., while the foreign companies are Google, Facebook, Microsoft, etc., and their translation technologies are getting better and better. In fact, the convergence of giants in the translation field is closely related to the evolution and commercialization potential of translation technology.

Technology is both hard power and principle

Google launched commercially deployable Neural machine Translation (NMT) in 2016 with an 86 percent accuracy rate, and announced the establishment of a Google AI center in Beijing, China, at the end of last year, which may soon see new achievements in its AI translation technology.

In China, Baidu launched the NMT system in baidu Translation APP in 2015, and showed a Simultaneous translation in Chinese and English at baidu World Conference in 2017. Robin Li also introduced that the accuracy rate could reach 95%, which shows that Baidu is also tireless in improving AI translation technology. Netease Youdao also launched NMT technology in May last year. Although it came late, it did not affect the reinforcement of its moat on the translation track.

And AI translation technology progress constantly learning, Google AI translation study process, for example, it is understood that the system USES the neural network of human supervision, the first thing you need to compare parallel texts, namely by human translation of books and articles before, in the text, and then by comparing the parallel large amounts of data, thereby learn peer relationship between any two specified languages, And gain the ability to switch quickly between them.

It can be seen that the evolution of AI translation technology is based on the premise of a large number of learning materials. In short, it can make up for what is missing. Good data materials, like high-quality education resources, can cultivate high-quality AI students.

The onslaught of giants combined with the constant innovation of NMT technology has become the norm in the field of AI translation. Depend on the result of a lot of artificial translation text, translation technology to be able continuously in learning to absorb and improve the accuracy in translation, such as tencent in a recent cooperation with Chinese and foreign translation company referred to in the current politics and professional class translation because are greatly influenced by the specific context, and AI has been eager to conquer the difficulties of translation techniques. The human translation of Chinese and foreign translation companies is like a nutrition source, constantly feeding Tencent’s machine translation technology, helping them to further improve and strengthen their learning ability.

It remains to be seen whether data-fed training will eventually allow Tencent’s AI translation technology to overtake on corners, with human translation with near 100% accuracy on one side and machine translation with imperfect skills on the other.

From the above cases, the evolution of translation capability is essentially the evolution of AI technology, feeding the technology with various translation data, so that its acquisition is closer and closer to the level of human translation. In general, in terms of technology, AI in the field of translation can represent a certain technical strength, which is also the direct reason why giants in the ERA of AI are willing to take root in this field. Once they have laid a solid technological foundation, they will be able to face the “mountain of swords and sea of fire” in the post-AI era.

The lure of commercial bounty

Giants are accustomed to commercial thinking, especially in today’s increasingly influential technological variables, while developing technologies, while implanting commercialized genes into products is widely recognized and commonly used product thinking. In certain ways, the value of a product must be measured in terms of money.

Therefore, the development of translation software, whether it is Google’s decision more than a decade ago, or Tencent’s sudden enlightenment, are difficult to escape from the “product commercialization” of this lesson. And this cultivation is bound to arise from the temptation of the commercialization of translation.

So far, Google, Baidu, Microsoft, etc. are conservative in the realization, or they do not aim for profit, positioning more in line with the free translation tools. Let’s take a closer look, Google, Microsoft, etc. do not even carry any ads, and Baidu only has half of the interface of the news stream ads on the home page.

However, this is only the tip of the iceberg, and the value transformation ability of translation technology has been quite amazing. For example, apps with online human translation services represented by Youdao Translator. , netease youdao announced last year that human translation orders exceed 1 million list, you can see the C end of the line and B end demand is strong, although the c-terminal translation needs is more dispersed, B side translation needs mainly focused on the line, but with the rise of the crowdsourcing model, online translation of professional translators and speed of response is as good as offline. In the long run, due to the optimization of online translation efficiency and results, there is no doubt that the demand for C-end translation will steadily increase. Driven by this part of user traffic and the advantage of online translation price, it is not unimaginable that the demand for B-end translation will see explosive growth in the future.

Clearly, the commercialization potential of translation technology has been proven by a number of companies, and everyone is still trying to tap into this huge gold mine. From Tencent’s recent actions with Chinese and foreign translation companies, we can also see that there are two blue ocean keywords in the field of translation. One is the continuous expansion of b-end demand, and the other is the continuous dividend of online translation. This can be interpreted as the new retailization of the translation market.

From another perspective, when many giants regard translation as a “public welfare” to do it, opportunities may come when you least expect them, because the user’s consumption upgrade if transferred to the field to the continued evolution of combining technology, imagine space is obviously a very large, then appeared, the scene is extremely likely giants “defect” commercial route, go back to the me.

Giant breakout method: new scene, second bonus, rigid entry point

In any case, the translation market is a giant training ground. Regardless of the order of entry, what everyone already knows is that this track will have greater possibilities and opportunities in the future. Now, in terms of the foresight and universality of AI itself, coupled with translation, which is the only way to overcome the communication barriers between people, players in the arena can break through and build a solid moat for themselves from several points.

First, consistent with the core idea of AI landing, casting a net on AI translation landing sets to grab new scenes. From baidu and other giants in the FREQUENT action on AI, no application, no business. Whether it’s Baidu announcing the advent of driverless cars, or Alibaba using AI to improve pig farming, AI won’t have any real value until it’s connected to real problems and needs.

AI translation is not about communication. Currently, there is a large demand for translation in tourism, trade, exhibition, education and other fields, so for the giants, spotting the landing opportunities in these fields is like grasping a small blue ocean, and even the early movers may use AI to create the Matthew Effect. After the conquest, the chain barriers built by AI translation may bring unimaginable development space for themselves and enjoy the benefits.

Second, use AI translation technology to explore traffic dividend twice. The advantage of the giant lies in flow, and flow is the basis of creating value, with flow, there will be a natural testing ground for new technology and new products. Facebook, for example, recently introduced new translation technology to cater to users’ needs for various conversation scenarios, and wechat has previously embedded similar translation functions.

For an international social application like Facebook, AI translation technology is a way to make the underlying needs explicit. If the results aren’t bad and users are willing to try out the technology, then a billion users is a low-hanging fruit for Facebook’s AI translation. The principle is not hard to understand. It’s just that the new feature satisfies the old requirement.

Turning to the giants on the floor, a similar scenario applies. For example, the traffic of Baidu’s search engine and Alibaba’s e-commerce shopping can match the supply and demand of AI translation technology to a certain extent. Moreover, the application of translation itself is also conducive to opening the door of internationalization for these giants and absorbing users from different countries.

Third, there are eye-catching rigid pointcuts. Google and Baidu also position themselves as free translation software for the public, which may be from the point of view of the public to meet the needs of users, and to a certain extent, this method also takes advantage of the user’s curious psychology of language, so there are considerable downloads. From the operation strategy, this can be understood as a low price (free) to seize the market behavior.

It is also a good choice to start from the vertical field, such as overseas translation officer for tourism, and Japanese, English and other translation software for language. At present, there are not only giants inside the market, but also many small and medium players. Therefore, vertical represents differentiation. If differentiation has lasting combat ability, such as the improvement of translation technology or the integration of added value, it may help players in the field to fight their way out.

One thing is for sure, players in the AI circuit must first follow a “technicality” approach before they can think about business. In the same way, translation cannot escape the fate of AI. Whether it is the giant who is high up and gobbling up food, or the small players who are treading on thin ice and thinking about things, they must first master the “raw material” of technology, and then build planes and cannons to break through the siege, or even to discover new lands.

Article/Liu Kuang public account, ID: Liukuang110