The top ten cutting-edge technology trends of artificial intelligence in 2024 were released in Beijing

The theme meeting of the 2024 World Science and Technology and Development Forum "Innovation in artificial intelligence governance to build an international trust foundation for cultivating the science and technology governance ecosystem" was held in Beijing on October 23. Qiao Hong, an academician of the Chinese Academy of Sciences and chairman of the World Robotics Cooperation Organization, released the "2024 Top Ten Cutting-edge Technology Trends in Artificial Intelligence Outlook" at the meeting, including 4 common AI technologies, 3 large-scale pre-trained models, 2 embodied intelligence, and 1 generative AI. It is hoped that it will trigger thinking and discussion among the academic community and the public, and jointly promote the development and application of artificial intelligence technology. The top ten cutting-edge technology trends of artificial intelligence released on the same day are: "The rise of small data and high-quality data", "Human-machine alignment: building a trustworthy AI system", "AI 'constitution': ensuring compliance and security", "Interpretable model: making AI more transparent and trustworthy", "Innovation of pre-trained models under the law of scale", "Full-modal large model: breaking data barriers", "A new era of AI-driven scientific research", "Embodied cerebellum model: giving robots real-time response capabilities", "Entity artificial intelligence system: wisdom empowers the physical world", and "World simulator: creating a digital world with infinite possibilities".

Among them, in terms of human-machine alignment, relying solely on data and algorithms is not enough to achieve human-machine alignment. It is necessary to transform human values and ethics into reinforcement learning reward functions to guide the model to learn behaviors that are more in line with human expectations.

The compliance, security and ethical issues of current AI systems are becoming more and more prominent, and it is particularly necessary to establish a supervisory model framework. Its main purpose is to ensure that all AI systems follow established principles during development and use by formulating clear standards and specifications, thereby reducing the risk of excessive use of AI without a certain system.

In terms of scale law, large-scale pre-trained models based on massive parameters and training data can effectively improve human-computer interaction and reasoning capabilities, and enhance the diversity and richness of tasks that can be completed.

In terms of AI-driven scientific research, large models, generative technologies, etc. are used to enhance and accelerate the efficiency of hypothesis proposal, experimental design, data analysis and other stages in scientific research, and improve research efficiency and accuracy. This flexible and efficient research method greatly increases the possibility of discovering new scientific laws, thereby accelerating the progress of scientific research.

In terms of the embodied cerebellum model, as an important regulatory center for robot movement, its core lies in solving the problem of combining software algorithms with physical space, as well as the contradiction between high performance of monomers and versatility of capabilities, so that intelligent robot systems can better meet the needs of fine operation and real-time control in the real world.

Humanoid robots are the ultimate manifestation of physical artificial intelligence systems. They not only have multimodal perception and understanding capabilities, can interact naturally with humans, but also can make autonomous decisions and actions in complex environments, and are expected to be applied to more complex work scenarios in the future.

The world simulator comprehensively considers data quality, diversity, model training strategies and regularization techniques, and the generalization ability of the generated model on unseen data is continuously improved. Combined with the digital interaction engine, it provides an immersive high-simulation experience, bringing users a richer and more diverse game world, which can be applied to education, entertainment and other fields, and can also create more super digital scenes.

Academician Qiao Hong pointed out that from intelligent manufacturing to smart cities, from medical health to financial services, the application of artificial intelligence is ubiquitous, and its influence is far-reaching and extensive, but it also faces new challenges and opportunities. How to grasp the development direction of artificial intelligence, how to promote technological innovation and industrial upgrading, and how to ensure the sustainable development of artificial intelligence technology, these issues need in-depth thinking and discussion.