Baidu Brain is the culmination of Baidu's years of technological accumulation and business practice, including core AI technologies and AI open platforms such as vision, speech, natural language processing, knowledge graph, deep learning, etc. It supports all Baidu's internal businesses and is fully open to the outside world, empowering partners and developers to accelerate the application of AI technology and enabling the transformation and upgrading of various industries. At the 2016 Baidu World Conference, Baidu Brain 1.0 completed the construction of basic capabilities and the initial opening of core technologies; at the 2017 Baidu AI Developer Conference, Baidu Brain 2.0 formed a complete technology system, opening up more than 60 AI capabilities; at the 2018 Baidu AI Developer Conference, Baidu Brain 3.0's core technological breakthrough was "multimodal deep semantic understanding," while opening up more than 110 core AI technology capabilities. Baidu Brain has opened up more than 150 leading AI capabilities to the public, building a full-stack AI technology layout. In the future, Baidu will continue to empower developers equally, allowing every developer to access AI capabilities in an equal and convenient manner.

Development History

With the search business as the core, Baidu was born as a search engine company in Zhongguancun in January 2000. Behind the search engine, in addition to internet technologies such as link analysis, AI technologies such as natural language processing and information retrieval are also needed. Therefore, Baidu began the research and application of artificial intelligence technology from its inception. Fully laying out the foundation of AI technology, in early 2010, Baidu, with 10 years of technological accumulation, began to fully layout artificial intelligence, successively starting the research and development of technologies including natural language processing, machine translation, speech, image, knowledge graph, machine learning, data mining, user understanding, etc. In January 2012, Baidu began to develop deep learning technology and launched voice recognition and image recognition capabilities that year, significantly improving recognition results. It was because of the stunning effects of deep learning in applications that Baidu established the world's first deep learning research institute in early 2013. In the second half of 2013, Baidu was the first in the world to apply deep learning technology to large-scale search ranking systems. In 2015, Baidu launched the world's first large-scale neural network machine translation system, more than a year earlier than Google. Opening up AI core capabilities to the outside world, at the Baidu World Conference in September 2016, based on Baidu's 16 years of deep accumulation, Baidu Brain was officially released, and it was announced that Baidu AI core technologies would be opened to the public. At the Baidu AI Developer Conference in July 2017, Baidu Brain 2.0 officially announced a complete technology layout including the foundational layer, perception layer, cognitive layer, and platform layer, and opened up more than 90 AI core capabilities through the AI open platform, including voice, image, video, augmented reality, natural language processing, etc., actively building the AI technology ecosystem together with industry partners and developers. At the Baidu AI Developer Conference in July 2018, Baidu Brain announced an upgrade to 3.0, opening up more than 110 core AI capabilities, and announced that Baidu Brain's daily call volume had exceeded 400 billion times [4]. The core technological breakthrough of Baidu Brain 3.0 is "multimodal deep semantic understanding," which not only allows machines to hear and see clearly but also to deeply understand the meaning behind them, thus better supporting various applications. On July 3, 2019, at the Baidu AI Developer Conference, Baidu CEO Robin Li began his speech. Robin Li revealed that Baidu Brain, as a technology platform with a full system of AI capabilities, has so far opened up more than 200 AI core capabilities to all developers. When these capabilities are combined with developers, they will produce magical power. "A single spark can start a prairie fire, and AI will definitely generate more and more value in various industries." [2] On September 15, 2020, at the Baidu World Conference live broadcast, Baidu CTO Wang Haifeng released the newly upgraded Baidu Brain 6.0 on site. According to the introduction, in terms of core technology, Baidu Brain 6.0 already has the capability of "knowledge-enhanced cross-modal deep semantic understanding." Based on the mastery of 550 billion pieces of knowledge, as well as the technology of "knowledge-enhanced continuous learning semantic understanding," Baidu Brain integrates different information such as voice, language, and vision to achieve cross-modal semantic understanding and obtain a unified cognition of the world. With this capability, machines can understand voice, recognize images and videos, understand language, and thus understand the real world. [3] On August 18, 2021, Baidu Brain 7.0 was officially released, and the upgraded Baidu Brain has two significant features: "integrated innovation and reduced thresholds."