IDC report issued jointly with the Baidu: forecast 2019 artificial intelligence Ten Trends

Recently, the International Data Corporation (IDC) and Baidu AI Industry Research Center (BACC) jointly issued a “Baidu brain Leadership White Paper”, the White Paper predicted 2019 China market trends of artificial intelligence, artificial intelligence, how to resolve the actual case from the technical to the floor , and that “100 days AI deployment plan.”
IDC expects, by 2022, China artificial intelligence market will reach $ 9.84 billion. From a technical white paper to, floor implementation, application value, ecological market predicts the four dimensions of artificial intelligence Ten Trends 2019: a prospect: Machine Learning / depth study into the traditional start of business. Machine Learning / depth study will enter the enterprise, to provide enterprises with the decision-making centered services. At the same time, the depth of learning will continue to widely used image, audio, text and other unstructured data processing. Especially in traditional industries in the medium-sized enterprises , the use of artificial intelligence, machine learning platform applications will become mainstream. IDC expects the industry before 2020, 15 percent of enterprises will use machine learning. Looking ahead: Fusion multimodal computer vision, speech, semantics has begun to drop. Multi-modal computing machine intelligence hear only be able to see has been unable to meet human needs, integration of visual, sound, meaning and emotion become an urgent need to achieve real intelligence just. The next three years will begin multi-modal computing landing in practical applications. Looking III: multi-model database began to move toward the market. With the investment in a business process of digital transformation of things, all kinds of unstructured data in the enterprise growth, making it possible to support a variety of multi-model database format data management is an urgent need. IDC estimated that in 2023, spending more than the model database will reach 30% NoSQL database expenditures. Looking in the fourth: low platform reduces the amount of code developed AI technology uses threshold. Low amount of code / code-free development platform to promote AI deployment automation, reducing the use of technical barriers, so that SMEs can use AI equality, achieve Pratt & Whitney AI. Users can upload pictures, audio, text and other raw data, the system can automatically train the appropriate model. A typical case – Google AutoML, Baidu EasyDL.
Looking V: AI is calculated from the cloud to the edge extension deployment. Infrastructure began to close to the data source and the end edge positions of the transfer devices, and artificial intelligenceIt will be the first to benefit from the application edge computing. Edge devices containing AI algorithms and drive computing power delivery. IDC expects 2022 Control Engineering Copyright , 25% of things end devices will run AI algorithm model. Outlook VI: business process intelligence, automation level reached new heights. Wave machine learning-driven artificial intelligence will promote a new round of business process reengineering of Control Engineering Copyright , many applications will be highly simplified. Typical case such as financial process automation, automated underwriting process automation level and many will reach new heights. IDC expects to 2023, artificial intelligence will replace 50% of the workload of the IT business, saving more than 20% of operating costs. Looking to seven: the trend of intelligent man-machine interface. On the one hand the speech-enabled programs more widely – spoken dialogue capabilities will be embedded into hardware and application software. On the other hand the integration of voice, image, video, and AI Semantic understanding will become the mainstream way humans interact with the application. IDC expects 2023 to support AI’s Human Interface will replace the current B2B 50% of screen-based and B2C applications. Looking to eight: six industries comprehensive use of AI. The government sector, the financial industry, the Internet industry after the practice in recent years through the application of comprehensive extended application of AI. The new retail, new manufacturing, medical field will also become a new growth point AI market. IDC expects the next three-year compound growth rate of the six industrial applications of AI will be more than 30%. Looking to nine: leading infrastructure software and applications. Software-defined computing has become an important strategy for chip manufacturers. AI driver software and application-specific chip phase will also come. Future, the popularity of the application of machine learning, machine learning whether always require large data sets, the evolution of the depth of learning neural networks CONTROL ENGINEERING China Copyright , like hardware accelerated computing will affect the development of the road. Machine learning technology evolution, AI application trend is increasingly important to the infrastructure provider. Looking ten: ecological resources integration as key to success. Artificial intelligence technology is penetrating to the end of the intelligent, highly successful adaptation of hardware and software applications can not be separated, which makes the integration of industry segments of technology-based companies with sensors, cameras, and other modules even more important. It can integrate solutionsVarious ecological elements of the case and build partner network platform has become the key to success.
So, how technology can be combined with practical application? IDC track after nearly 70 scenarios found that, with the opening of the market, more and more rich technical capabilities, extensive application scenarios of trends, and want artificial intelligence technology floor to enterprise applications and exerts efficacy, efforts need three points: First, difficulty of use, simple operation technology stack, two vertical depth of the scene is based on tuning of the model, the third is from the data center to the end of the software and hardware adaptation. White Paper proposes a preliminary framework to assess the effectiveness of AI application, the AI ​​will affect the sector into goods and services, production model, business model, decision-making model four categories CONTROL ENGINEERING China Copyright , and evaluated from various dimensions. For example, in manufacturing, artificial intelligence will give priority to goods and services, production patterns and to bring high-performance business model; in the financial industry, artificial intelligence application performance best reflected in goods and services as well as operational intelligence links, followed by the production of smart and intelligent decision-making. Artificial intelligence can bring a different performance for different business sectors, but almost all of the cases brought by artificial intelligence are reflected in the performance of saving time, labor and other productive resources, reduce costs and enhance productivity, income growth and so on.
The White Paper also proposed that artificial intelligence is experiencing ecological changes from “highly concentrated” to “fragmented” and then to “carry out their duties,” the high degree of integration, a clear division of labor upcoming AI ecology. In this trend, for the full application performance, industry participants should not stop at the mere adoption of these technologies, but should actively build and enable the parties to help promote cooperation platforms and services, so as to enhance the efficiency of the entire ecosystem.
White Paper stressed that with the continuous improvement of artificial intelligence in the industry penetration, qualified enterprises need to develop artificial intelligence deployment plan as soon as possible. IDC recommends that enterprises planning to deploy AI 100 days plan, step by step to complete the plan, select the scene, set up a subcommittee, testing, implementation, replication, this cycle continually adjusted to accommodate a wider range of business needs.

Related articles