【Commentary】Writing poems, paintings, making medical diagnoses... Recently, the conversational large-scale language model technology represented by ChatGPT and MOSS has attracted widespread attention. On March 3, Qiu Xipeng, a professor at Fudan University and a professor at the AI Model Algorithm Research Center of Pazhou Laboratory (Huangpu), said at the award ceremony of the first Guangdong-Hong Kong-Macao Greater Bay Area (Huangpu) International Algorithm Research Competition held in Guangzhou that the emergence of large-scale language models will accelerate the realization of general artificial intelligence.
Not long ago, Qiu Xipeng's team released China's first large-scale language model MOSS. Qiu Xipeng said that compared with ChatGPT, the scale of MOSS parameters at this stage is relatively small, and the cost of research and development is relatively low, which can be used in vertical industries with a small amount of knowledge and only about 10 billion parameters. However, due to its small capacity and limited knowledge capacity, it still needs to be improved in terms of answering questions and logical reasoning ability, and then the knowledge in this field can be expanded by expanding the scale of parameters or accessing the knowledge base of the vertical field to expand the knowledge and enhance the reasoning ability.
【During the same period】Professor Qiu Xipeng of Fudan University
For ChatGPT, I think the current level of it may be that we can catch up, or it will not be too difficult to achieve, but it is also developing. So if we want to catch up, or catch up with it, we have to put in more effort.
These may require efforts in several aspects, and I think the most important thing may be that our ecology or mechanism can encourage innovation and (establish) our scientific research or innovation mechanism. Our advantage in China is that in fact, we have a large number of talents, and there are a large number of application scenarios, so we will give full play to these advantages, and then driven, for example, and then the underlying algorithm innovation, can further catch up.
Overseas leading research institutions have recently begun to turn to general artificial intelligence based on large models. Qiu Xipeng explained that the ultimate goal of achieving the development of artificial intelligence is to achieve general artificial intelligence, which is the same as the image of artificial intelligence in science fiction movies. He said that based on the capabilities currently on display, we are not far from the arrival of the era of general artificial intelligence.
【During the same period】Professor Qiu Xipeng of Fudan University
That is to say, we are optimistic that we can achieve the general artificial intelligence we have been pursuing in the near future.
Then the future form may be similar to what we in science fiction movies, such as giving it an anthropomorphic image, and then giving it emotions, or a certain sense of self.
Zhu Jun, a professor at Tsinghua University and a professor at the AI Model Algorithm Research Center of Pazhou Laboratory (Huangpu), said that China's artificial intelligence market is active and rich in scenarios, and it is necessary to establish Chinese large-scale language models.
【During the same period】Professor Zhu Jun of Tsinghua University
Our artificial intelligence market in China is still relatively active, and this scenario is very rich, because our language is mainly Chinese, so I think for China, there must be large-scale pre-trained models like Chinese or Chinese (Chinese large-scale language models). In fact, now that everyone has taken an inventory, it basically means that we have also built a lot of this (artificial intelligence computing center) in China, and we are now on this platform, we can do new model algorithm innovation.
It should be pointed out that the road of artificial intelligence must go far, and it is necessary to establish red line constraints. Qiu Xipeng suggested that artificial intelligence products should be restricted by laws and regulations and some algorithms should be eliminated through the model side to remove harmful information.
【During the same period】Professor Qiu Xipeng of Fudan University
In the future development of artificial intelligence models or algorithms, ethical issues must be very important. So in this process, we must strengthen ethical supervision, then I think this supervision can be (started) from many levels, when we input training data to the model, we must do a good job of data governance in advance, that is, to "depoison" it, remove the harmful information in it.
Reporting by reporters Cai Minjie and He Junjie in Guangzhou
Responsible Editor: [Song Fangcan]