In the blink of an eye, 2022 has passed. This year, the COVID-19 epidemic has brought many challenges to the global economy and social life, and the characteristics of extreme uncertainty have become more obvious. At the same time, the power of science and technology continues to emerge, the speed of technological progress and industrial development is further accelerated, and digital technology and intelligent technology bring more imagination space to people's lives.
In today's era of rapid change and full of unknowns, how to explore more certain values and directions? Intelligent technology forms the main line of today's technological changes, but in which specific areas will AI technology bring more surprises in the foreseeable future?
At the beginning of the new year, Baidu Research Institute released the "Top Ten Technological Trend Forecasts in 2023", which pointed out the main line of technological development in the era of "technology to reality" and elaborated on the ten major technological trends in three dimensions. Major specific trends:
# 1. The technical base is continuously consolidated. The deep learning platform enlarges the model and builds a solid industrial intelligence base, which will further accelerate the upgrading of industrial intelligence. The integration of data and reality provides strong impetus and broad market space for consolidating the technical base. Privacy computing has become an important cornerstone to support data security governance and the market-oriented development of data elements.
#2. AI cross-domain integration is more solid. The development of AI technology brings huge value space for cross-field and cross-disciplinary integration innovation. Cross-field integrated innovation represented by Al for Science is developing a complete tool system, transforming ideas into practical values, and promoting the transformation of scientific research paradigms and the development of emerging industries. AI-powered robots are increasingly used in work scenarios that require a large amount of manpower to improve labor productivity.
#3. Intelligent innovation is more pragmatic. Based on a solid intelligent base, on the one hand, the digital economy and the real economy accelerate the integration and promote the upgrading of the industrial system; on the other hand, web 3.0, metaverse, autonomous driving, AIGC, quantum New directions and fields such as science and technology will accelerate development and implementation.
AI large models are evolving towards cross-language, cross-task, and cross-modal technical directions, and have become an important factor in the development of current AI technology. trend. Relying on the deep learning platform, the performance of large model technology has been continuously improved. It has the advantages of strong versatility, generalization, interpretability, and high degree of standardization of the development process. It can solve the problem of AI fragmentation and continue to reduce the complexity of AI development and application threshold.
As large model technology gradually matures, training capabilities, core operator libraries and software platform layouts continue to improve. In aerospace, finance, energy and other fields, "industry large models" have begun to emerge. , build AI infrastructure around the needs of various industries and promote application innovation in the "AI industry". We predict that in 2023, large industry models will cover more fields, gradually form an ecosystem, actively implement "inclusive AI", and serve the intelligent upgrade of thousands of industries.
my country's "14th Five-Year Plan" and the 2030 Vision Goals outline both emphasize the vigorous development of the digital economy, which has brought strong innovation power and broad market space to digital technologies such as artificial intelligence. At present, new AI infrastructure such as intelligent computing centers, deep learning platforms and large models are constantly being consolidated to accelerate the industrialization of artificial intelligence and meet the needs of the transformation of the real economy, which is dominated by manufacturing; at the same time, my country's huge manufacturing industry, rich Application scenarios and massive data resources are very conducive to the iterative evolution of deep learning models. The integration of technology and scenarios will give birth to new products and new business formats with industrial versatility.
AI new infrastructure construction, in the short term, will become one of the important tools for local governments to develop the digital economy, and will have a significant role in promoting regional economic development and industrial upgrading; in the medium and long term It will promote the deep integration of digital technology and the real economy and better empower the digital transformation and intelligent upgrading of industries.
People are building a content-rich virtual world. This virtual world was originally parallel to the physical world and gradually developed into a closely connected physical world. In the future, it is possible to achieve mutual interaction, integration and symbiosis. This trend is expected to accelerate in 2023, thanks to breakthroughs in numerous key digital technologies.
Web 3.0 technology will create a decentralized, more open, fair and secure new network space where users can exchange information and value more securely; digital twins and virtual simulations The integration of , AI and other technologies will create more accurate and intelligent digital twins in many scenarios such as shopping, manufacturing, home, and cities; AIGC will bring a new content creation model and be widely used in fields such as painting, literature, and even video production. , significantly reducing content production costs; combined with the immersive experience services brought by VR/AR and the high-speed big data transmission capabilities provided by 5G, a comprehensive and rich Metaverse business format is expected to accelerate the formation of products and service models based on this. It may also usher in a new round of disruptive innovation.
As autonomous driving enters urban scenes, the difficulty of perceiving complex environments and processing massive data has greatly increased. Traditional small models cannot meet the requirements of high-level autonomous driving. The industry has begun to introduce large model technology to allow autonomous vehicles to effectively expand semantic recognition data, greatly improve the efficiency of solving long-tail problems, further enhance the generalization ability of autonomous driving perception, and adapt to more travel scenarios.
It is expected that in 2023, the commercialization of autonomous driving in major cities in China will show a trend of double growth in operation scope and fleet size, and the market entry rate of smart cars with autonomous driving technology will also increase. With new breakthroughs, the smart car industry has moved from "testing the waters and sailing" to "sailing the long voyage".
## As the world ages, various industries will face severe labor shortages in the future. Many countries have actively developed automation technology to change the dilemma of labor shortage.
The maturity of key technologies such as artificial intelligence, big data, and cloud computing has injected strong impetus into the development of automation. In particular, various types of robots powered by AI will play an important role in real-time perception, intelligent decision-making, and Optimization control and other aspects have been greatly improved, and it is increasingly used in construction, mining, disaster relief and other work scenarios that require a large amount of manpower. In addition, humanoid robot products will play the role of housekeepers in life, undertaking simple carrying, cleaning, nursing and other tasks, which will not only greatly improve labor productivity, but also free people from heavy physical labor to a certain extent, allowing people to have more Spend more time enjoying the good life.
By introducing AI technology, researchers have developed scientific computing tools to solve problems that are too complex and difficult to solve in traditional scientific computing and improve system modeling and analysis capabilities. It is believed that more powerful scientific computing tools will appear in the future, promoting AI technology to become an important auxiliary force for scientific research, reflecting its unique value in basic sciences such as physics, chemistry, biology, materials science, and drug research and development and other application fields.
In the past year In 2018, quantum computing technology has achieved a new round of breakthroughs in key technical directions such as software and hardware, applications, and networks. It is expected that in 2023, the performance indicators of quantum chips using various technical routes will continue to improve. Cloud-native quantum computing platforms will provide more powerful, richer, and more professional services. The ease of use will be greatly improved, and the development threshold will be further reduced. Quantum devices will be used in It has shown advantages in multiple application scenarios, and more quantum algorithms with practical application value have been born in the directions of artificial intelligence, material simulation, financial technology, biopharmaceuticals, etc.; with the continuous improvement of quantum computing hardware performance and quantum algorithms, quantum software and hardware The value and demand for integrated solutions will become more prominent. The public's attention to quantum information science will continue to increase, and there will be more extensive and urgent needs for quantum science popularization and education. With the help of quantum computing platforms, we will accelerate the formation of a prosperous quantum computing industry ecosystem. , the multi-party collaboration and resource pooling between government, industry, academia and research will be further deepened. The importance and urgency of data security governance and marketization of data elements are increasing day by day. Privacy computing technology has entered a stage of rapid development. More and more institutions in finance, communications, medical, Internet and other fields have begun to Self-built privacy computing platforms, application scenarios continue to expand and deepen, and promoting the interconnection of various privacy computing platforms has gradually become a new trend in the industry. In this context, a "horizontally and vertically intertwined" trusted data circulation network has initially emerged. It is foreseeable that with the help of the continuously growing data circulation network, the application scenarios of privacy computing technology will continue to be innovative in the next few years, and the privacy computing platform will also become an important cornerstone in supporting data security governance and the market-oriented development of data elements in many industries. Help shape a data industry that takes into account both value creation and security and trustworthiness. The rapid development of artificial intelligence and other technologies has brought new social ethical issues and risks, attracting the attention of countries around the world. This year, the Chinese government issued the "Opinions on Strengthening the Ethical Governance of Science and Technology" and submitted the "Position Paper on Strengthening the Ethical Governance of Artificial Intelligence" to the United Nations, actively advocating the principle of "people-oriented, intelligence for good" to ensure the safety, reliability and controllability of artificial intelligence. Technology companies and scientists are also actively exploring explainable AI technology, trying to promote effective human-machine communication in the context of value alignment, so that AI can truly understand human intentions and reduce the " Black box risk" to achieve more predictable AI governance. Good technology not only focuses on results, but also on process. We predict that in a highly intelligent and digital society in the future, having trustworthy and controllable AI technology capabilities will become a new competitive advantage for enterprises. In recent years, under the influence of the concept of sustainable development, promoting energy conservation, emission reduction, cost reduction and efficiency improvement has become an important evolution direction of new technologies. Among them, edge computing takes into account the real-time and elasticity of computing, can reduce the transmission of massive data, and save huge data transmission and energy costs. In the future, edge computing will collaborate with 5G, AI and other technologies to help the development of a low-carbon economy. Advanced computing is improving the scale of existing computing power, reducing computing power costs, and improving computing power utilization efficiency from multiple levels such as computing theory, architecture, and systems. We predict that there will be more new technological breakthroughs focusing on green, low-carbon and sustainable development capabilities in the future, and their implementation is expected to alleviate issues such as environmental protection, health, energy and materials, and improve The quality of human living environment. Nowadays, AI technology has a more effective development direction, and pre-trained large models have become a new technical fulcrum; intelligent The effect of AI technology is becoming more and more significant. From industry to scientific research, the value space of large-scale implementation of AI technology has been revealed, and this is also Baidu's main consideration in predicting technology development trends. Looking back in 2022, Baidu has made phased breakthroughs in three major directions and fields: large industry models, quantum computing and virtual-real symbiosis, especially in industrialization. With results. 1. Take the lead in launching large-scale industry models and verify the feasible path to industrialization of large-scale models Baidu will launch a large-scale model in the industry in May 2022 In September, we were the first to launch an industry large model. With "industry knowledge enhancement" as its core feature, the Wenxin large model is based on general data training, plus mining of a large number of industry characteristic data and knowledge existing in industry application scenarios, and combined with discussions with industry experts. , introducing sample data and unique knowledge accumulated from actual business in the industry, designing characteristic algorithm tasks in the industry field, and improving the adaptability of large models to industry applications. As of November 2022, Wenxin has released a total of 11 large industry models, covering electric power, gas, finance, aerospace, media, cities, film and television, manufacturing, social science and other fields , achieving significant effect improvement, the industrial ecosystem has initially taken shape, and is becoming an important force in accelerating the intelligent transformation and upgrading of the industry. At present, the Wenxin large model has been widely used in Internet products such as search, information flow, and smart speakers, and empowers industries such as industry, energy, finance, communications, media, and education through the Feipiao open source open platform and Baidu Intelligent Cloud. Baidu’s industrial practice has proven that cooperating with industry leading companies to develop industry large models is an effective way to implement the large model industry, allowing large models to be used in actual application scenarios Not only is it usable, it’s also convenient and easy to use. 2. Let quantum move from the laboratory to industrialization and lay the foundation for the quantumization of the digital economy 2022 In August, Baidu released the superconducting quantum computer "Qianshi" and the world's first full-platform quantum software and hardware integrated solution "Liangxi", opening up the entire chain from hardware, platform to application, realizing quantum chips to quantum applications The transformation takes quantum from the laboratory to industrialization and lays a solid foundation for the quantization of the digital economy. Baidu superconducting quantum computer "Qianshi" integrates quantum hardware, quantum software, and quantum applications. It is an industrial-grade superconducting quantum computer Among them, the quantum hardware platform of Qianshi Superconducting Quantum Computer is now equipped with a 10-qubit high-fidelity superconducting quantum chip to provide users with stable and high-quality quantum computing services; Qianshi Quantum Software The platform supports obtaining quantum computing power through cloud services and provides industrial-level quantum computing services. Liangxi provides a series of services such as privatized deployment, cloud services, and hardware access. It has the ability to adapt to many types of mainstream quantum chips such as superconducting and ion traps, and can realize quantum chips instantly. Plug and play, users only need to download an APP to experience quantum computing, which simplifies the entire process of deploying quantum hardware to quantum services to the greatest extent. 3. Virtual and real coexist, AIGC’s commercial value has been verified by many parties With the prosperity and development of digital technology, the virtual world The connection with the physical world is getting closer, and technologies such as Web3.0, Metaverse, and AIGC continue to expand to a wider group of people. In 2022, as an innovative method of content generation, AIGC is realizing large-scale applications in stimulating creative inspiration, improving content diversity, and reducing content production costs. The AIGC capabilities of Baidu Wenxin large models have been demonstrated on multiple occasions: #On October 20, the "People's Daily" released a video of the future China picture painted by AI based on some words in the 20th National Congress report. The video was provided by Baidu Feiping , Wen Xin Yige provides technical support. Looking at the history of human science and technology development, driving the scientific and technological revolution and industry The core technologies of change often have strong versatility. When it has the characteristics of standardization, automation and modularization of industrial mass production, the value effect of core technology as infrastructure will become more and more "real". In this process, the application threshold of technology is continuously lowered and the application effect is continuously enhanced, ultimately supporting the magnificent industrial upgrading and social progress. There is no doubt that today’s artificial intelligence is at such a stage. This year is the fourth consecutive year that Baidu Research Institute has released annual technology trend forecasts. Over the past four years, the technological landscape is always changing, opportunities continue to emerge, and the tracks are constantly changing. It can be seen that a number of technological trend predictions previously released by Baidu Research Institute have become the true trajectory of subsequent technological and industrial development. For example, digital people and virtual people will appear in large numbers and shine in the service industry; in interdisciplinary and cross-field research, especially in the fields of quantum, biology, chemistry and other fields, "AI" will become one of the most exciting topics. Research direction... Facts have proved that Baidu Research Institute’s precise insights into cutting-edge technologies and market patterns have become a benchmark for the development of related fields and industries. The annual publication of annual technology trend forecasts also fully demonstrates Baidu's mission as a pioneer and practitioner of innovative technologies such as artificial intelligence and quantum computing. Wang Haifeng, Chief Technology Officer of Baidu and Dean of Baidu Research Institute, said: "Because we have seen the irreversibility of the trend of intelligence and verified the inevitability of AI becoming real. This year's top ten technology predictions We will continue to continue this line of thinking, hoping to help clarify the future technological development context and anchor the direction of technological innovation." "Looking for a clear direction for AI and technology to become real, is Baidu Research Institute shares with you the original intention of predicting technology trends in 2023. I hope we will take this opportunity to find the right direction, promote sustainable development, and work tirelessly to achieve high-level technological self-reliance and self-reliance." The best way to predict the future is to create it. Seeing this, are you ready for the explosion of these intelligent technologies? AI is moving towards reality and technology towards reality, anchoring the general direction of scientific and technological innovation
Baidu annual science and technology trend forecast, using the certainty of science and technology to grasp the uncertainty of the future
The above is the detailed content of Baidu Research Institute releases the top ten technology trends in 2023, anchoring 'AI towards reality': the industry's large model ecology is emerging, and intelligent innovations such as autonomous driving, AIGC, and quantum technology are more pragmatic. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
