The unforgettable 2020 has passed, and the new year still has great uncertainty, but the trend of technological revolution and industrial transformation led by information technology is even more obvious. What will happen to our technology in 2021? What impact will these changes have on our social economy?
On January 13, Baidu Research Institute officially released the top ten technology trends forecast for 2021, covering many frontier fields such as artificial intelligence, digital humans, life sciences, transportation industry, AI chips, Internet of Things, quantum computing and so on.
Reading through the "Baidu Research Institute's Top Ten Technology Trend Forecasts in 2021", it is found that this trend forecast is very technologically forward-looking, fully demonstrating Baidu's leadership in basic technology research, such as multi-modal AI, unsupervised learning, Kunlun Chips, quantum computing, etc. At the same time, since each item of the forecast trend is closely related to the industry and is not a pure technical document, it has extremely high reference value for guiding the development of the industry.
Compared with the science and technology forecast trends issued by Dharma Academy and Tencent Research Institute, Baidu Research Institute, which has been deeply involved in AI for many years, has significantly more predictions on AI.
Xi Ge talked about his views on the development of science and technology in 2021 according to the "Baidu Research Institute's Top Ten Science and Technology Trends Forecast in 2021", especially the impact of AI on us.
01
In 2021, industrial AI will further deepen its development, and AI will go deeper into people’s lives
In "Baidu Research Institute's Top Ten Technology Trends Forecast in 2021", each item is closely related to AI. Let's take a look at the first four prediction judgments.
1. The epidemic accelerates the integration of AI, and AI will penetrate deeper into the lives of the people.
2. Digital humans and virtual humans will appear in large quantities and shine in the service industry.
3. Unsupervised learning becomes the basic support technology, promotes the integration of AI multi-field technology, and enhances the ability of cross-modal understanding.
4. The transportation industry is accelerating its transformation, and "digital transportation operations" will become a key driving force.
Judging from the prediction of Baidu Research Institute, in 2021, AI will go deeper into various industries and become an important technology for improving efficiency. At the same time, AI's impact on people's lives will be more specific.
Baidu Research Institute predicts that "AI will go deeper into the lives of the people" is based on AI's active performance in fighting the epidemic and resuming production in the past year. During the 2020 epidemic, a camera system composed of infrared cameras and face recognition cameras created by Baidu, an algorithm system independently developed by Baidu, and an AI multi-human body temperature rapid detection solution composed of a computer acting as a computing power and display system The plan has effectively improved the scientificity and effectiveness of epidemic prevention and control at multiple front-line epidemic prevention gateways such as aviation, high-speed rail, highways, and stations. The AI temperature measurement equipment provided by Baidu, based on the detection of key points of the face and the image infrared temperature dot matrix temperature analysis algorithm, quickly screens and warns the forehead temperature of multiple people in the crowded area, and solves the problem of fewer facial recognition features caused by wearing masks and hats It is convenient to quickly screen people gathering places, and improve detection efficiency and accuracy.
In addition, in order to solve the problem of crowd control and investigation in community prevention and control, Baidu's AI free smart outbound platform is also playing a huge role. This platform provides outbound services in three scenarios: mobile personnel investigation, local residents' return visits, and notification of specific groups of people. It assists grassroots communities in epidemic investigation and quickly finds out health and mobility conditions. The efficiency is hundreds of times higher than that of manual phones.
But as the Baidu Research Institute said, AI will further integrate with basic industries such as transportation, energy, manufacturing, and agriculture to promote the resumption of work and production, and promote social and economic development. At the same time, remote office, online education, online consultation and other modes will continue and become the new normal in the future.
"Robotaxi vehicles without safety officers carry out normal operation." Brother Xi still remembers the exclamation of the on-site personnel when Apollo completed the world's first fully driverless live broadcast debut (without safety officers). But as Baidu Research Institute said, AI and 5G technologies will be deeply integrated with transportation and become a new engine for transportation development. The normalized operation of Robotaxi vehicles without safety officers is expected to become a reality, and the "smart cockpit" featuring intelligent voice interaction and vehicle-road collaborative information services will become the mainstream standard for automobiles.
"Digital people", Baidu Smart Cloud has taken the lead in the industry in this regard. Baidu Smart Cloud and Shanghai Pudong Development Bank jointly launched Xiaopu, the first "financial digital person" in China in 2019. In 2020, Baidu also launched Du Xiaoxiao, a virtual AI assistant for mobile phones.
In terms of industrial AI, Baidu Smart Cloud, as the carrier and exporter of AI to B, is exporting Baidu AI technology to smart cities, smart finance, smart medical, smart manufacturing, smart energy and other fields. For example, smart manufacturing covers 14 major industries, more than 100 companies, more than 30 partners, and more than 50 types of vertical scenes. It has been implemented in 3C, automotive, steel, energy and other industries.
There is no doubt that in 2021, the impact of AI on the industry and on people's lives will be further enhanced.
02
In 2021, cutting-edge technologies such as biological computing, AI chips, and quantum computing will become hot spots
"Baidu Research Institute's Top Ten Technological Trends Forecast in 2021" pointed out that biological computing, AI chips, and quantum computing will be hot spots in 2021. This is also the forecast content of 5, 6, 7, and 8.
Baidu Research Institute's 4 predictions on the development trend of cutting-edge technology in 2021:
1. Life science has become a new territory for AI applications, and biological computing has become a new research hotspot.
2. AI chips and lightweight models accelerate the penetration of intelligence to the edge, and the development of the Internet of Things usher in new opportunities.
3. AI improves the machine's understanding of reality, and service robots will have more physical interaction capabilities.
4. The concept of quantum technology innovation is deeply rooted in the hearts of the people and directly promotes the development of quantum computing related industries.
Xi Ge was not surprised that Baidu Research Institute included biocomputing as one of the top ten trends in 2021, because it played a huge role in the anti-epidemic and epidemic prevention in 2020.
In 2020, in order to support the fight against the epidemic, Baidu Research Institute announced that it will open the linear time algorithm LinearFold and the world's fastest RNA structure prediction website for free to various genetic testing institutions, epidemic prevention centers and scientific research centers around the world to enhance the new coronavirus RNA spatial structure prediction speed. According to media reports, with the help of LinearFold technology, scientists have shortened the whole genome secondary structure prediction of the new coronavirus from 55 minutes to 27 seconds, which is 120 times faster. At the same time, Baidu also announced that it is committed to scientific research on coronavirus research. Institutions, fully open artificial intelligence technology support and billion-level computing supporting resources. Baidu launched the world's first mRNA vaccine gene sequence design algorithm LinearDesign, which can greatly improve the stability of vaccine design and protein expression level within 16 minutes, thereby effectively solving the most important stability problem in mRNA vaccine development and improving vaccine development speed. In the 14th Protein Structure Prediction Competition (CASP14) held in 2020, Google’s AlphaFold2 achieved atomic or close to atomic structure prediction for most proteins, leaving academic predictions far behind, shocking The entire computational biology and structural biology community. As Baidu said, in 2021, the combination of AI, biocomputing and virus research, vaccine research and development and other basic research fields of life sciences will usher in an explosion.
"AI will improve the machine's understanding of reality, and service robots will have more physical interaction capabilities". In fact, most of this has been achieved. In December 2020, Baidu released a new robotic innovation technology called "TransFormer with Visual Tokens for Human-Robot Interaction, TFVT-HRI" (TransFormer with Visual Tokens for Human-Robot Interaction, referred to as TFVT-HRI). This technology is the industry's first attempt to extend active interaction to a multi-modal interaction mode that integrates expressions, actions, and rich speech skills. Through the target detector, the relevant objects that may have an impact on the active interaction are extracted into visual tokens covering vision and relative spatial information, and then the Transformer network is used to learn the relationship between visual tokens over a period of time, so as to achieve Spatiotemporal modeling of the interactive agent to predict whether the interactive agent has the potential to interact. In other words, robots using this technology have a more specific understanding of reality, are more intelligent, humane, and have more powerful functions. If you can complete some daily tasks in the home scene, such as helping the elderly get drinks from the refrigerator, organizing clothes, etc. As Baidu said, there will be more hotel application service robots in the future to respond to room demand in real time and automatically deliver items. The tasks that home service robots can complete will be more sophisticated and diverse, such as helping the elderly to take food or brewing drinks from the refrigerator, while avoiding collisions with other objects in the house or moving people, and can also help organize the room when the owner is out. Wait.
03
The application of technology accelerates, and technological ethics has become a new social issue
It has to be said that the application of new technologies allows us to enjoy numerous technological benefits. Application service robots to make our lives more convenient, temperature AI, AI interrogation techniques applied to enhance our ability to respond to outbreaks, and application Linearfold Lineardesign technology has accelerated our vaccine development speed ...... we While enjoying the convenience brought by new technologies, the ethics of technology application has also become a problem we must pay attention to.
The 9th and 10th articles of "Baidu Research Institute's 2021 Top Ten Science and Technology Trends Predictions" reflect on this aspect.
Baidu Research Institute's prediction on AI ethics and talent training in 2021:
1. The society's demand for building responsible AI continues to rise, and technology companies promote implementation.
2. Lower the threshold and develop talents simultaneously, and the popularity of AI accelerates.
Baidu Research Institute believes that AI ethics will promote the development of smart applications from capturing user attention to promoting user well-being. In 2021, from the perspective of technology landing, the interpretability and robustness of AI models will make more progress, and the dispute between the "spear" (synthesis) and "shield" (identification) of video synthesis technology will intensify. Privacy protection computing will also accelerate its use among organizations, providing security and privacy guarantees for unlocking the value of data. At the same time, the rapid development of deep learning in recent years has also greatly increased energy consumption. In the future, the development of AI will pay more attention to selective computing and pursue sustainable economic, social and environmental development.
Xi Ge believes that Baidu Research Institute's thinking on responsible AI reflects Baidu's in-depth thinking on technology ethics and technological temperature, and is a concrete manifestation of the company's practice of social responsibility.
Seeing the prediction of Baidu Research Institute on "lowering the threshold and training of talents, and accelerating the popularity of AI", Xi Ge thought of some of Baidu's initiatives this year. On June 19, 2020, Baidu announced that it will continue to increase its investment in new infrastructure areas such as artificial intelligence, chips, cloud computing, and data centers in the next ten years, and announced the "two 5 million" plan. Specifically, by 2030, the number of Baidu Smart Cloud servers will be expanded to 5 million units-this number is equivalent to 50% of global server shipments in 2019, which is the total computing power of TOP500 supercomputers currently available in the world. 7 times. In addition, in order to ensure the sustainable development of the entire AI research field, Baidu will help the industry cultivate a 5 million-scale AI talent team in the next ten years, and provide strong talent support for the full implementation of the new infrastructure. On July 2, Baidu announced that it will train 50,000 AI data annotators at Baidu's Shanxi Data Annotation Base in the next five years and introduce more AI partners.
As Baidu said, in the future, the large-scale application of AI and the demand for greater computing power will drive a new round of growth in AI chips, help AI break through the current computing power bottleneck, and release AI computing power more inclusively. High-quality data can effectively improve the ability to understand application scenarios. Customized data solutions, optimized data collection and labeling, and a sound data element market system will promote the application of AI in various segments. In addition, the popularization of AI is inseparable from talents, whether it is research-oriented, application-oriented talents, or the cultivation of AI literacy is indispensable. I believe that as technology companies further promote the integration of AI production and education, more industries will emerge. Many industrial compound talents who understand both AI technology and AI scenarios.
Xi Ge believes that the "Baidu Research Institute's Top Ten Technological Trends Prediction in 2021" is a condensate of Baidu AI's research results, and it is Baidu's experience summary and predictive discussion of many years of AI industry practice.
In 2021, which is still full of great uncertainty, the release of this document provides a reference development direction for people's digital, networked and intelligent development of the economy and society in the post-epidemic era of global economic recovery.
Understand the "Baidu Research Institute's Top Ten Technological Trends Predictions in 2021", he will be your guide to "see through" the future and find certainty from uncertainty.
PS: It is recommended that interested friends view the original Baidu "Baidu Research Institute's Top Ten Technology Trends in 2021".