SMEs embrace the correct posture of artificial intelligence

When it comes to data, there is no doubt that it is an important asset of this era. The data reflects the principles and laws of things. When you find its rules, you can predict the unknown. If the data is crude oil, then AI (Artificial Intelligence) is a processing plant that extracts various high-value products from crude oil. Its importance is evident.
Discovering knowledge, insights, and patterns from data is not a new concept in and of itself. Hundreds of years ago, there was such practice in the Kepler era. At that time, Kepler extracted and summarized the three laws of celestial movement from hundreds of pages of celestial position data, which is still being used today, which is known as the Kepler's three laws. Now, AI helps us realize the automatic learning of knowledge and rules from massive amounts of data by means of large-scale cloud computing.
So, what role can it bring to us as a data-driven AI framework?
First, a data-driven AI framework can bring a personalized experience. For example, when we enter some websites, we get a lot of personalized experiences. These experiences make the site no longer a thousand people, and every user can make adjustments and optimizations through the data-driven AI framework. Effective personalized service comes from in-depth analysis of large amounts of data, and AI helps us accurately match the most appropriate experience to each user.
Second, the data-driven AI framework can bring a fine-grained industry strategy that can help companies operate refineably. For example, a target customer base for a product can be roughly defined as a male or female of a certain age range. After applying the data-driven AI framework, we can get a more detailed description. We can not only consider factors such as age and gender, but also cross-consider more dimensions, such as hobbies, behaviors, etc., to get fine-grained. Marketing strategy.
Finally, data-driven AI frameworks can bring knowledge and insight. We can learn new knowledge from experience, and the core value that the data-driven AI framework brings to us is the ability to continually and operationally mine knowledge and learn from data. This knowledge is not necessarily written in textbooks, but from the data in real-time, maximum mass and at the same time the most effective acquisition of knowledge for production and business practice. Similarly, continuous insights can be obtained from the data through the AI.
One of the cores of AI is de-averaging. For example, for a company, the average value of a customer may be one hundred yuan, and de-averaging tells us that the value of different customers is different. This can be learned from AI, from past customer behavior data and other attributes, to establish a self-learning model to predict the value of each customer. The actual value of the customer may be a far cry from the average. Not only the value of the customer, whether the customer will buy a product, what kind of product they like, and how to make a purchase can be answered by AI technology. De-averaging applications are not limited to marketing but can be applied to medical and other business areas. For example, in the medical field, the probability of scurvy based on cases and the probability of re-admission can help hospitals save patients and reduce medical costs. These applications have already begun in some large hospitals.
The famous science and technology thinker Kevin Kelly said that AI is cognitive. If electrification brings artificial motivation, then cognitiveization brings artificial intelligence. A lot of practice shows that in terms of perception, including AI's visual, auditory, and language understanding, AI can approach the human brain; in support of professional decision-making, AI can even surpass the human brain with the support of massive data.
There are many applications such as this, the core capabilities of Data + AI for us to build a developing enterprise service ecosystem, including industry applications, such as finance, medical, education, etc.; in addition, there are cross-dimensions in each industry, It is a functional application, such as marketing, customer relationship management, security and other related functions. Industry applications and functional applications form a two-dimensional matrix, and AI has many application scenarios in it.
Practice tells us that AI large-scale commercial application scenarios should have two necessary conditions: 1. The quality and quantity of data must meet certain requirements, especially the opening of the entire data flow and regular data updates, which determines whether the basis of AI development is Firm; 2. There is a clear definition of the problem in the field. If the domain itself does not have a clear definition of the problem, it is difficult to solve the problem through AI. From an industry perspective, finance has come closer to these two points; from a functional perspective, some highly digitized industries such as marketing, customer relationship management, and security are relatively close.
As a growing company, embracing AI will face some challenges. In the past ten years of practice, we have found some common challenges.
First, companies must fully understand the value of data. Many companies want to use big data to drive business growth, but data resources are scarce like other high-quality resources. From the outset, companies need to design specific business, product, and technology architectures to ensure that ongoing operations can precipitate relevant data. For companies with a certain customer volume, valuable data may already be in your database. Companies that don't realize this or don't know how to mine the value of data will miss the opportunity to use AI.
Second, it is imperative to discover and cultivate AI-related talents. As we all know, the scarcity of data scientists has a great impact on this emerging field. In foreign countries, training programs such as InsightData Science have had a positive impact on the industry. As the ranks of data scientists grow, data scientists working in AI in businesses are becoming more common. In contrast, I think the industry is even more lacking is the AI ​​product manager. In communication with many companies at home and abroad, one of the complexities of the AI ​​problem is the uncertainty of the results, and there are very few product managers with AI background, which can not judge the value and direction well, and lead to related products or projects. Shelving. Of course, there are also problems in talent cultivation. For example, we can try to encourage excellent data scientists and engineers to lead the development of related products. Under the guidance of the business, give full play to the enthusiasm of professional talents and explore feasible directions.
Third, the integration and integration of cross-disciplinary teams is the key to landing. After the closed loop of data is opened, the close integration of products, engineering, and AI often takes a long time to run in. For example, in our experience of building an AI platform, it involves a lot of team communication, cooperation, and mutual support. Whether AI can be effectively realized and the construction of engineering capabilities is particularly important. Because AI is not only an algorithmic problem, it is difficult to implement continuous, large-scale AI applications in big data without a strong data processing infrastructure. Based on this need, the team of pure engineers and the team of pure scientists often fail to help AI to the most effective. Only the deep integration of the team and the business can create greater value.
Fourth, if AI wants to make value for industry users, it must solve the problem of trust. As an emerging way of thinking and technology system, AI is a common bottleneck in the process of solving practical problems in the industry. This trust includes trust in data and algorithms. A certain range of data sharing can increase trust, which can bring new knowledge and insight. In addition, in the landing scenario within the enterprise, it is also important for the builders and users of AI to establish trust. This requires regular and effective communication between teams based on the results of backtesting or measurement.
In different industries and functions, AI, whether it is corporate decision makers or executives, will face a variety of problems, including some common challenges. If these challenges are solved, I believe that not only large enterprises, but also small and medium-sized enterprises will have a relatively large space to use AI upgrades - using best practices in the AI ​​field, conducting rapid proof of concept, and falling production on the premise of risk control.
Ding Lei | Wen Dinglei is the chief data scientist of Baidu Finance. He has served as the CTO of Huibaichuan Credit Information and the head of the Global Consumer Data Science Department of PayPal.

Indoor Lighting

Indoor Lighting,Interior Lights,Wall Lights Indoor,Led Interior Lights

Foshan Nanhai Pinming Electrical Appliance Co.,Ltd , https://www.mavcadonf.com