Ai has undoubtedly become a hot investment area, but practitioners who are committed to the deep application of AI in the vertical industry need to remain calm and avoid being unduly disturbed by the external environment. Artificial Intelligence is a variety of frontier science of comprehensive disciplines, its complexity is visible, throughout the development of artificial intelligence, but also after years of accumulation, until the cloud computing, big data to obtain large-scale application, the popularity of the speed can be improved. If we divide artificial intelligence into two paths of scientific research and engineering application, then we can see more progress in scientific research, and the speed of actual application is behind scientific research. Why is it? I think it's systemic and complex. Memories of our college days, we also contributed to scientific research, not only improved a lot of algorithms, but also can output scientific research papers, and eggs, we all know that the armchair can not solve real life problems. in the face of a real problem domain, we can have a large number of solutions to choose, how to select the optimal solution, which is in itself complex. In the case of Amazon Go's unmanned store, many people say there are no new technologies, either electronic tags, or computer vision, or scanning or NFC. In fact, when we think about one step further, we know the difficulty. For example, electronic tags, in a closed system, electronic tags are able to solve the object recognition of one of the schemes. However, it would not be a good idea to apply an electronic tag to the actual storefront product. To the market for the circulation of goods to increase the electronic label, either in the store when the goods are made by the shop, or in the factory when the product is done by manufacturers. By the store to do, increased labor, and can only be affixed to the packaging of goods, can not deal with the product shelves after the consumer tear, change, demolition, in the unmanned shop will increase the loss. By the commodity manufacturers do certainly better, but manufacturers to do for the business, to a certain extent also increased production costs, and the general business is difficult to promote the upstream supply chain to do this thing. In the case of computer vision, the simple image of the object recognition technology is still difficult to achieve a commercially accurate rate, not to mention the object recognition in the video. Faced with a large number of goods in the store and the camera in all directions, how to cross-check the real-time calculation of the consumer's products are what kind of, its calculation efficiency, recognition accuracy and other aspects of the requirements are very complex. One more step, even if every single technology is feasible, when integrating these technologies together into an unmanned store system, is more than simply a plus and a relationship? Let's think about the evaluation of the application effect of artificial intelligence. When artificial intelligence technology is applied to a field, how to evaluate its effectiveness can prove the effectiveness of artificial intelligence? There is still a lot of space for research in this area. Still taking Amazon Go's unmanned store as an example, how do you evaluate the results of using these AI technologies? Is the cost reduced? It doesn't look like a shop assistant, but it also adds a lot of cost to sensors, equipment and technology. It seems that consumers are more efficient, but what is the direct result of AI? When it comes to using AI as a solution scenario, the scenario itself is already complex enough that it is difficult to simply attribute the post-application results to the success of artificial intelligence. So, for the engineering application of artificial intelligence, can not be carried away because of the concept of fiery, think that artificial intelligence to solve everything. In the long run, AI is still a solution to many problems, but the evaluation and proof of this effect is not overnight, and it is not possible to arbitrarily negate the role of artificial intelligence because of the inability to give quantitative evaluation simply. Conversely, it may be more convincing to focus on the trend effect of applying AI in the process of its functioning. In the early stage of AI application, as long as the application of artificial intelligence in the technical solution can be used to speak, logical, or can be bold attempt to explore while improving. Of course, it takes time to prepare for the long term.
--Welcome reprint, please specify the original source http://blog.csdn.net/caowenbin - -
The cold thinking of artificial intelligence
