Face-to-face (facing the multiple challenges of generative artificial intelligence like ChatGPT, how

1. Artificial intelligence chatGPT4

Recently, Fudan University held a special report meeting, where Zhao Xing, Deputy Director of the National Intelligent Evaluation and Governance Experimental Base and Professor of the Big Data Research Institute at Fudan University, and Jiang Zhuoren, a researcher at the "Hundred Talents Plan" in the Information Resource Management Department of the School of Public Management at Zhejiang University, discussed the challenges and governance of generative artificial intelligence online and offline.

2. Artificial Intelligence ChatGPT Download

From Siri voice assistant, which was first launched by Apple mobile phone in 2011, to Everything Everywhere All at Once, which won seven Oscar awards this year, generative AI has already been integrated into human life. ChatGPT tools have further ignited the global AI boom. Zhao Xing and Jiang Zhuoren together bring an academic feast that runs through the past, present and future of generative AI, and analyze the multiple challenges posed by generative AI.

3. How to download chatgpt

What is Generative Artificial Intelligence? At present, there is no unified definition for generative artificial intelligence, which can generally be understood as an artificial intelligence system that can generate text, images, or other media information based on prompts. "This technology does not appear out of thin air like magic, but has a long history." Jiang Zhuoren emphasized that in 1932, French engineer Georges Artsouni created the device "mechanical brain".

4. Artificial Intelligence ChatGPT Concept

It completes translation by querying a multi-functional dictionary. The input and output are all a strip of paper. Jiang Zhuoren believes that although it is different from today's machine translation, it fully conforms to today's definition of generative AI, that is, human inputs a piece of content, and machines generate a new content. In the 20th century, MIT created Eliza, one of the earliest generative AI, Judea Pearl introduced the concept of Bayesian network causal analysis, and Yann Lecun and others showed how to use convolutional neural networks to identify images

5.chatGPT

In 2006, Chinese computer scientist Li Feifei began building a database called ImageNet, which contained over 14 million manually annotated images and over 20000 categories. "It is precisely with the support of such a huge database that deep learning can rise. Therefore, we always say that good research taste, coupled with continuous efforts, can lead to a good academic career.

6. How to read chatgpt

”How did Jiang Zhuoren evaluate ChatGPT as a celebrity product? With a parameter count of 175 billion and a training word count of 300 billion, ChatGPT has attracted over 100 million users in just two months since its release in 2022, making it the fastest growing application in history.

7. Microsoft's Artificial Intelligence ChatGPT

Jiang Zhuoren believes that in order to understand ChatGPT, it is necessary to understand its key technologies: large model basic training, instruction fine-tuning, and human feedback reinforcement learning. Faced with the challenges brought by generative artificial intelligence, Zhao Xing interprets it from four dimensions: resources, technology, application, and social ethics. From the resource dimension, generative artificial intelligence requires high-quality data, while the data quality in the Chinese world is weaker than that in English.

8. The hottest artificial intelligence ChatGPT

Zhao Xing believes that even with strong translation capabilities, the Chinese processing performance of ChatGPT like tools is significantly weaker than that of English. One of the core reasons is the poor quality of the Chinese data fed. In addition, hardware technology is also a key factor supporting the rapid development of China's artificial intelligence industry. "When we are preparing to launch a universal tool into the whole society, but cannot clarify its scientific principles, there will inevitably be endogenous risks.

9. Strongest Artificial Intelligence ChatGPT

”The core point of artificial intelligence risks lies in the unbearable consequences. "We are rarely in such a weak state in governance issues," Zhao Xing said. At the application level, the certainty of the development of generative artificial intelligence industry and the uncertainty of risk governance will persist for a long time. In society, generative artificial intelligence not only deeply falls into the problem of intellectual property rights and information leakage, but also shapes a true information cocoon.

10. Artificial Intelligence ChatGPT Composition

When generative artificial intelligence is around you 24 hours a day, subtly, you will mistakenly believe that everything is your own decision, "Zhao Xing warns." We are facing something that will rise in a short period of time or cause serious consequences with unknown consequences. "" Can we find a way to resist unknown risks before they erupt? This is the problem that generative artificial intelligence endogenous security governance needs to solve.

”Zhao Xing said, "We need to light up a new skill tree for human society before the risks of artificial intelligence come: the ability to deal with non-traditional security issues in artificial intelligence.

When we know what potential risks are and where they arise, generative artificial intelligence governance has the opportunity to transform into routine security issues, and we can try to find the implementation of closed-loop governance, "Zhao Xing said." However, this still requires long-term exploration in theory and practice. "Zhao Xing's team began to explore the application of generative artificial intelligence in scientific evaluation, We have innovatively constructed a new paradigm for the harmonious coexistence of "objective data, intelligent algorithms, and expert evaluation" in the evaluation and governance of "digital intelligence individuals".

Recently, the team is also conducting exploratory experiments on using ChatGPT like tools to construct an intelligent evaluation system. Xinmin Evening News reporter Zhang Jiongqiang