Just after 2019, artificial intelligence is continuously advancing along a stable development trajectory. What can these give us? In what direction should we focus in 2010? To this end, Baidu Research Institute recently released a forecast of ten major technology trends in 2020.
In the past 2019, we have seen many developments in the field of artificial intelligence:
At the technical level, the emergence of tools such as AutoML has lowered the technical threshold for deep learning. At the hardware level, the emergence of various AI-specific chips provides computing power for large-scale applications of deep learning. In addition to AI, the development of Internet of Things, quantum computing, 5G and other related technologies has also provided many conveniences for the penetration of deep learning in the industry.
The rapid development of these underlying technologies and the increasing trend of technology integration means that in 2020, we are about to enter the era of “industrialized” mass production of AI.
Recently, based on the development of artificial intelligence technology in the past decade and the practical experience of industrial applications worldwide, the Baidu Research Institute officially released the 2020 top ten technology trends forecast.
The following is a detailed interpretation of the top ten forecasting trends:
Trend 1: AI technology has reached the stage of industrialization, and there will be many “AI factories” in 2020
AI technology itself and various business solutions have matured and are rapidly entering the “industrialization” stage. With the continuous investment of AI technology from domestic and foreign technology giants, there will be many AI model factories and AI data factories around the world in 2020. AI technologies and business solutions will be produced on a large scale and used in various industries to help the industry. upgrade. For example, AI solutions in the customer service industry can be replicated and applied to industries such as finance, e-commerce, and education on a large scale.
Trend 2: 2020 will be a critical year for the large-scale landing of AI chips
In recent years, AI chips have gradually reached a usable state, and 2020 will be a key year for the large-scale landing of AI chips. End-side AI chips will be more low-cost, professional, and integrated solutions. At the same time, the NPU (Neural Network Processing Unit) will become the basic module of the next generation of general-purpose CPU chips. In the future, more and more end-side CPU chips will use deep learning as the core for new chip planning. In addition to the chip, AI will also redefine the computer architecture, supporting AI training and predictive computing as a new heterogeneous design architecture idea.
Trend 3: Deep learning technology penetrates the industry and is applied on a large scale
Deep learning is currently the most important field in artificial intelligence and the most effective technology proven by the industry. An open source deep learning platform with a deep learning framework at the core has greatly reduced the threshold for the development of artificial intelligence technology and effectively improved the quality and efficiency of artificial intelligence applications. In 2020, all walks of life will apply deep learning technologies on a large scale to implement innovations and accelerate transformation and upgrading.
Trend 4: Automatic machine learning AutoML will greatly reduce the threshold of machine learning
AutoML will be able to integrate the iterative processes in traditional machine learning to build an automated process. Researchers only need to input meta-knowledge (such as the convolution operation process, description of the problem, etc.), and the algorithm can automatically select the appropriate data, automatically optimize the model structure and configuration, automatically train the model, and adapt it to deployment To different devices. The rapid development of AutoML will greatly reduce the threshold of machine learning and expand the penetration rate of AI applications.
Trend 5: Multi-modal deep semantic understanding is further mature and more widely used
Multi-modal deep semantic understanding takes different modal information such as sound, image, text as input, and integrates AI technologies such as perception and cognition to achieve multi-dimensional deep understanding of information. With the rapid development and large-scale application of technologies such as vision, speech, natural language understanding, and knowledge atlas, multi-modal deep semantic understanding has further matured and application scenarios have become wider. Combined with AI chips, it will be widely used in the Internet, smart home, finance, security, education, medical and other industries.
Trend 6: Natural language processing technology will be deeply integrated with knowledge, and computing platforms for general natural language understanding will be widely used
With the emergence and development of large-scale language model pre-training technology, the general natural language understanding ability has been greatly improved. The pre-training technology of semantic representation based on massive text data will be deeply integrated with domain knowledge to continuously improve the effectiveness of natural language processing tasks such as automatic question answering, sentiment analysis, reading comprehension, language inference, and information extraction. A universal natural language comprehension computing platform that combines ultra-large-scale computing power, rich domain data, pre-trained models, and improved research and development tools will gradually mature and be widely used in the Internet, medical, legal, and financial fields.
Trend 7: IoT will make breakthroughs in three directions: boundary, dimension and scenario
With the development of 5G and edge computing, computing power will break through the boundaries of cloud computing centers and spread to all things, creating a pan-distributed computing platform. At the same time, time and space are the two most important dimensions of this physical world, and insights into time and space will become the basic capabilities of a new generation of IoT platforms. This will also promote the integration of the Internet of Things with more scenarios such as energy, power, industry, logistics, medical care, and smart cities, creating greater value.
Trend 8: Intelligent transportation will accelerate its landing in diverse scenarios such as parks and cities
The development of autonomous driving is becoming more rational, and the market will be more confident about the development of intelligent driving in the next few years. In 2020, more autonomous vehicles will be used in different scenarios such as logistics and express delivery, public transportation, and closed roads. At the same time, V2X (vehicle to everything) technology started large-scale deployment and application, making vehicles, roads widely connected, further promoting the realization of intelligent vehicle-road collaboration technology, and intelligent transportation accelerated in diverse scenarios such as parks, cities, and high speed Landing.
Trend 9: Blockchain technology will be integrated into more scenarios with a more pragmatic attitude
With the deep integration of blockchain technology with AI, big data, IOT, and edge computing, the data and offline mapping problems are solved one by one. Solutions such as data confirmation, data use, data circulation and exchange built around the blockchain will play a huge role in all walks of life. For example, in the field of e-commerce, the authenticity of the data of the entire process of the goods can be guaranteed; in the field of the supply chain, the openness and transparency of the entire process of data can be guaranteed, and the secure exchange between enterprises, the realization of electronic documents and so on.
Trend 10: Quantum computing will usher in a new round of explosion, injecting new vitality into AI and cloud computing
With the successful demonstration of “quantum hegemony”, quantum computing will usher in a new explosion in 2020. In terms of quantum hardware, the performance of programmable medium-scale noisy quantum devices will be further improved and it will initially have error correction capabilities. It will eventually be able to run quantum algorithms with a certain practical value, and quantum artificial intelligence applications will also be greatly development of. In terms of quantum software, high-quality quantum computing platforms and software will emerge and achieve deep integration with AI and cloud computing technologies. In addition, with the initial formation of the quantum computing ecological industrial chain, quantum computing will definitely get more attention in more application fields. More and more industry giants are investing in R & D resources for strategic layout. Have the opportunity to bring a new look to the future of AI and cloud computing.