What Is ChatGPT and Its Industrial Applications?

ChatGPT is a new chatbot model developed by the AI research lab OpenAI, using the Transformer network architecture, and released on November 30, 2022. It is an AI technology-driven natural language processing tool.

It can engage in conversations by learning and understanding human language, and interact based on the context of the chat, truly conversing like a human. It can even complete tasks such as writing emails, video scripts, copy, translations, and coding.

ChatGPT is the chatbot model launched by OpenAI on November 30, 2022, and its web application allows users to use it for free. Currently, interactions between users and ChatGPT include casual chats, information inquiries, poetry and essay writing, code modification, and more.

ChatGPT Chat Example

2022 can be called the inaugural year of AIGC (Generative AI). In the spring of 2023, driven by ChatGPT, a new wave of AIGC is bound to rise.

Just one month after ChatGPT was launched, Google CEO Sundar Pichai issued a “Code Red” internally, asking multiple teams to focus on addressing the threat posed by ChatGPT to the company’s search engine business, while also approving plans to integrate AI chatbots into Google Search.

Microsoft views ChatGPT as a new generation of technological revolution and announced additional investments in ChatGPT developer OpenAI, integrating its AI tools into a series of products including Bing search engine, Office suite, Azure cloud services, and Teams chat program. Microsoft plans to invest $10 billion to acquire a 49% stake in OpenAI (negotiations are currently ongoing), prompting Google to invest over $2 billion in ChatGPT competitor and AI startup Anthropic to counter Microsoft. Baidu plans to launch a generative search based on ChatGPT in March.

How to Integrate with Industrial Applications?

He Xiaodong, Vice President of JD Group and IEEE Fellow, is optimistic about AI development and believes that AI will inevitably follow an industrial route, with more opportunities for application in industry than in academia.

He Xiaodong stated that from the user experience, ChatGPT feels refreshing, both in fluency and logic, but the problem is that the content is relatively low in “substance”, with stronger logic than content. From a technical perspective, the biggest innovation of ChatGPT lies in the new semi-supervised learning algorithm, which allows it to understand user intentions, potentially reaching the level of large-scale commercial search engines.

He Xiaodong believes that compared to the previous extensive use of unsupervised deep learning algorithms, the algorithms and training processes behind the ChatGPT model are more novel. Without human data selection, even if the model parameters are significantly larger, it would be difficult to achieve this effect. Especially, ChatGPT has made the generated text model more “organized”, which is a significant technological innovation. “In a sense, this is actually a correction of the previous blind pursuit of (parameter) size and unsupervised learning,” He Xiaodong stated.

ChatGPT Q&A Example

Huang Minlie, a tenured associate professor in the Department of Computer Science and Technology at Huazhong University, mentioned that after GPT-3, OpenAI has not open-sourced any of its models, but it provides API calls. In this process, it has done something significant: establishing a real user call and model iteration flywheel, emphasizing the importance of real-world data calls and their impact on model iteration. Of course, during this process, it has also supported a large number of startups in the U.S., establishing an ecosystem.

He Baohong, director of the Cloud Computing and Big Data Research Institute of the China Academy of Information and Communications Technology, believes that in the digital native era, it is impossible for everyone to understand programming, thus low-code and no-code are necessary. Traditional low-code and no-code platforms mainly rely on visualization and modularization, while in the next 10 to 20 years, AIGC will enable non-programmer creators to use natural language commands for zero-error software development. Low-code/No-code 2.0 should be based on AIGC, truly merging business and technology, making programming more accessible.

Specifically, what application scenarios does ChatGPT have in industrial manufacturing? A technical expert from Kunlun Data Zhihui Data Technology summarized:

• 1. Cross-industry Science Popularization: ChatGPT can answer various questions related to technology and industry, quickly simplifying complex technical knowledge into easily understandable forms, thus helping people outside the professional field understand it.

• 2. Engineering Technical Support: ChatGPT can help engineers find code examples and technical solutions more quickly, accelerating the development process. For example, generating code based on requirements provided by engineers or offering code modification suggestions, while reminding best practices and common errors during the writing process.

• 3. Knowledge Updating: ChatGPT can help employees in industrial companies quickly learn and update technical knowledge, such as providing the latest technical materials and research progress based on requirements given by engineers.

• 4. Data Analysis: ChatGPT can analyze large amounts of data to help industrial companies better identify patterns and trends, and make decisions based on this.

Q: What else can be done with secondary development based on ChatGPT in industry?

An expert from Kunlun Data K2 believes: “Currently, ChatGPT has reached a cautiously usable level in some references for code and text content, but it still lacks the depth for data value mining. This depends on the training data and model configuration of the ChatGPT model. If the model has been trained and the training data contains relevant industrial data, it may be able to perform data analysis. However, if the model has not been trained or the training data is insufficient, it may not meet the needs.

Moreover, the model’s configuration is also crucial, as it will affect the model’s accuracy and efficiency. How the model is trained, how it is configured reasonably, and how to obtain more diverse real data to adequately nourish the model, I believe there is still a long way for ChatGPT in the industry.”

Who Will ChatGPT Replace?

Recently, the United Nations Conference on Trade and Development (UNCTAD) published an article on its official website titled “How AI Chatbot ChatGPT Affects Employment and Jobs”. The author, Shamika N. Sirimanne, Director of the Technology and Logistics Division of the organization, believes: “Like most technological revolutions that impact the workplace, chatbots have the potential to create winners and losers and will affect both blue-collar and white-collar workers.” Tools like ChatGPT also pose a real risk of job loss for technical and semi-skilled workers.

Sirimanne wrote in the article that, for example, chatbots can be developed to train employees within organizations, leading to redundancy of human trainers.

In recent years, several new jobs related to cutting-edge technologies have emerged, including social media managers and drone pilots, etc. However, these adaptations take time, and the breakthrough changes that cheap AI personal assistants can bring may significantly affect office workers.

Moreover, the impact of this technology is not limited to developed economies. Sirimanne believes in the article that with gig work and digital transformation, there is an expectation that some technical workers in developing countries will be ready to compete for more skill-intensive jobs worldwide as “remote immigrants”, such as accountants, legal clerks, software developers, and even X-ray analysts. The risk is that these jobs will be taken by ChatGPT from around the globe.