But why wasn't ChatGPT born in China? Experts interpret it this way: the "Hundred Model Battle" is becoming increasingly intense as it unfolds
Artificial intelligence is affecting our lives at an unprecedented speed. Since the beginning of this year, the research and development of large models at home and abroad has shown explosive growth. AI technology represented by large models is accelerating iteration and upgrading, reshaping thousands of industries. Large models have also become a high-frequency term at this year's Bund Conference.
What is the current development status of the generative AI industry represented by large models? What new opportunities have new technologies brought? On September 7th, at the "New Wave of AI Frontier Technology Development and Application" Insights Forum, attending experts shared their views.
"Since ChatGPT, big models have become particularly popular," said Zhu Wenwu, a professor at the Department of Computer Science and Technology at Tsinghua University and director of the Big Data Intelligence Center at the Institute of Artificial Intelligence. The characteristics of big models can be summarized in several "big" terms: "big" models, "big" data, and "big" calculations. "We need big models, big data, and big computing to make big models particularly good."
The development of China's large model industry is in full swing, and many industry insiders describe it as a "hundred model war".
At the Bund Conference, AI painting tools only need to provide a creative idea, which can be combined with personal favorite art styles through text or images.
However, the "heat" of domestic large-scale models has sparked Liu Zhiyuan, an associate professor at the Department of Computer Science and Technology at Tsinghua University, to think "coldly": "After ChatGPT was released, we have conducted many discussions. Why was ChatGPT not born in China?"
In Liu Zhiyuan's view, on the one hand, China's AI industry has formed a plateau, but has not yet reached its peak. At present, there is a "hundred model war" in the industry, showing an inward trend and lacking in original innovation. However, "I think we should be delighted with this industry ecosystem, because from a global perspective, China should be the place that fully embraces the big model. We have a very good plateau, and we will definitely have our own peak in the future."
The "Hundred Model Battle" is becoming increasingly intense, which has also sparked reflection among industry insiders. In the view of Xu Peng, Vice President of Ant Group, the current development of large models still lacks application. "To influence the development of big models and AI, we still need to return to applications." He said that diversity of applications can bring value to people and society, which in turn can have more motivation to invest in supporting big models and basic research, turning the plateau into a peak.
The attending experts also discussed the future development trends of large models. In Zhu Wenwu's view, maintaining emergence ability, coordinating large and small models, and lightweight vertical domain large models are the main directions for future research.
Liu Zhiyuan believes that big models are the technology that moves towards universal artificial intelligence. "The future development path of big models includes tool learning, multimodality, embodied intelligence, swarm intelligence, etc. For example, we need to fully utilize the language interaction ability of big models and try to construct a 'think tank'. This is one of the future development directions." Liu Zhiyuan said that big model technology will help us better embrace the technological revolution of artificial intelligence, allowing humans to better understand and transform the world, and have a better society.