"The goodness of ChatGPT is scary. We are not far from an AI that is powerful to an extent that it can be dangerous." This was the feeling expressed by Elon Musk after using ChatGPT. Even Microsoft’s co-founder Bill Gates said, "The significance of this technology is no less than the birth of the Internet or personal computers." ChatGPT, an AI chatbot, was launched by OpenAI's AI research lab at the end of November 2022, and quickly got popularity on social media. In just 5 days, the number of registered users exceeded 1million, and just two months later, this number surpassed 100 million making it the fastest-growing consumer application in history. OpenAI was founded in 2015 by serial entrepreneur Elon Musk, Y Combinator’s President Altman, PayPal’s co-founder Peter Thiel, and others, aiming at creating "universal" robots and chatbots that use natural languages.
In the past decade, artificial intelligence has gradually become a hot cake in the social and economic fields. Artificial intelligence (AI for abbreviation) is a new technology science that studies and develops theoretical, methodological, technical and application systems for simulating, extending, and expanding human intelligence. On July 8, 2017, the State Council of China issued and implemented the " Development Plan for the New Generation of Artificial Intelligence" to grab the strategic opportunities of AI technology, to build China's key advantages in AI development to accelerate the construction of an innovative country with a world technology power. In the past decade, the application and development of artificial intelligence have made a remarkable progress, with new technologies constantly emerging, which has attracted strong attention from industries and the public. For example, the facial recognition technology brings more convenience to smartphone users; . the appearance of AlphaGo makes people exclaim that human intelligence has been replaced by machines; the emergence of autonomous driving technology makes people think that they can free their hands while driving, and so on. However, these artificial intelligence technologies often quickly lose their popularity from the public attention. But, why has ChatGPT continued to be so popular?
Is ChatGPT just a process of quantitative change to qualitative version? In 2018, OpenAI released GPT-1 with 117 million parameters; in 2019, OpenAI released a model GPT-2 with 1.5 billion parameters; in 2020, OpenAI released the latest GPT-3 model with 175 billion parameters; in March 2022, OpenAI released their InstructGPT model based on the GPT-3 model and further fine-tuned it. From theaforementioned iteration of different versions, it is obvious that OpenAI has increased the number of parameters from 117 million to 175 billion in the latest generation. Thus, is it just a process of quantitative change? In fact, as early as 2017, the Google Brain team published a paper entitled "Attention is all you need" at the Neural Information Processing Systems Conference. The authors proposed a transformer model based on self-attention mechanism, and applied it to understand human languages for the first time i.e. natural language processing. It is this paper that revised the defects of the mainstream model of recurrent neural network (RNN) in natural language processing. The RNN model has problems with unstable models or premature termination of effective training while dealing with longer sequences, such as long articles or books, and it also takes a long time to train the model. The Transformer model proposed in 2017, can perform data computation and model training in parallel which takes less training time. The trained model can also be grammatically interpretated, which means that the model is interpretable. Obviously, ChatGPT is not just a process of quantitative to qualitative change, but a product with both a "miracle of power" (quantitative change) and a "miracle of brain" (qualitative change).
Is ChatGPT merely a process of information to knowledge? Many AI products, such as image recognition, speech recognition, machine translation, etc., predominantly recognize and convert "existing" information for users, rather than providing "non-existing" knowledge to users. Although previous searching engines could provide users with “existing” information that they didn’t have, what they provided are still “information” rather than “knowledge.” Users need to read, sort out, and process this information to develop their own “knowledge,” which requires plenty of time and experience. Obviously, the emergence of ChatGPT has solved the pain points that many users could not solve when using search engines, and created a large number of justifiable demands. Nevertheless, ChatGPT's current capabilities still have many drawbacks, for example many users have found that many of ChatGPT's answers are "correct but senseless talks," even "pretentious nonsense," and some answers are too "slick". This requires ChatGPT to continuously improve its ability to answer questions through constant knowledge learning, moving towards answering more “objective” questions in the future. Clearly, ChatGPT is not only a process of "providing information" (miracle of power), but also a process from "learning knowledge" to "outputting knowledge" (miracle of brain). Of course, we hope that it can move not just towards "outputting knowledge", but to "outputting ideas" as well, and even to "creating innovations" in the future.
The development of human history is a process of constantly liberating "powers", from the "physical power" to "computing power" and then to "brain power". In the liberation process, humans constantly innovate, which continuously drives the development of economy and civilization in human society. Obviously, ChatGPT is also part of the constant “power liberation” process in the human history. With the machine's "power", it is undeniable that many of the "powers" owned by humans are unable to compare with the “powers” of machines. However, we need have no fear; we should face the future optimistically and make an reasonable use of machine's "power" and "brain" to create greater "innovations" for humanity.
The utilization of the "power" of machines is to create greater "innovation" miracles for humanity. From using animal power to steam engines, to internal combustion engines, to electric generators and to nuclear power, humans have mastered the"engine power" to an increasingly greater extent and our capabilities have become stronger and stronger. With the invention of abacuses, slide rules, calculators, and computers, humans have also progressively mastered the "computing power." Such powers have allowed humans to break through many obstacles that were previously perceived as unbreakable. The powers allow us to go further, climb higher, calculate faster and more accurately. Therefore, how to make the worthy use of "power" of machines, and generate even greater "innovative" miracles for humanity, has become the new opportunities along with many challenges that humanity faces today.
The utilization of the "brain" of machines is to create greater "innovation" miracles for humanity. Humans have been using computers and algorithms to develop machine’s "intelligence" from “Deep Blue” computer to the “Deeper Blue” one, to AlphaGo and to ChatGPT. According to the study by a professor at Stanford University, ChatGPT has already possessed the intelligence of a 9-year-old child. It is reasonable to predict that ChatGPT will possess the intelligence of an adult even surpass it in the near future. Therefore, it has brought new opportunities and challenges for human beings to make a good use of machine’s brainpower to create even greater "innovation" miracles.
From the emergence of ChatGPT to its continuous development, it is evident that "innovation" is at the core of the success. It is an important aspect to foster students’ innovation spirit in the education process in China. system. If we say education in the past merely focused on "knowledge transmission" through structured methods, "knowledge-breaking" will become a core component of education in the future. We believe that the utilization of the "power" and "brain" of machines provides new "power" for humans to break through the boundaries of existing knowledge. Furthermore, it is more important to combine the "power" and "brain" of machines with human ingenuity to achieve new disruptive "innovations" and continuously create new "miracles".
——Hang WeiProfessor,Associate Executive Dean College of Business, Shanghai University of Finance and Economics