Overview of ChatGPT profile, impact of future development and concerns raised

Published By: EAIOT Time: Apr 21, 2023 11:40:26 Categories: ChatGPT 833 Views Total: 0Comments

Artificial intelligence is rapidly evolving and I believe the biggest buzz this year has been ChatGPT. Its influence has hit the entire tech scene. People are discussing how to apply ChatGPT to solve problems in their lives, while at the same time discussing the impact of AI on their lives and the issue of privacy and security. So what is ChatGPT? What kind of development has it undergone? Let's go and learn more about it!


I. Overview of ChatGPT

ChatGPT (full name: Chat Generative Pre-trained Transformer), a chatbot program developed by OpenAI, was released on November 30, 2022. It also interacts with the context of the chat, truly communicating like a human, and can even perform tasks such as writing emails, video scripts, copywriting, translating, coding, and writing essays.

In April 2023, the China Payment Clearing Association (CPCA) advocated for payment industry practitioners to use ChatGPT with caution.

On 12 April 2023, Italy, which had previously temporarily banned the use of ChatGPT, offered to lift the ban, but on the 13th, Spain's National Data Protection Authority and France's National Commission for Freedom of Information each announced an investigation into ChatGPT. On the same day, the European Data Protection Board announced the establishment of a special working group to facilitate collaborative national investigations and to exchange information on possible enforcement measures to be taken by the respective data protection ministries.


chatgpt.jpg


II. History of ChatGPT's development

ChatGPT is a new artificial intelligence technology-driven natural language processing tool from OpenAI, an American artificial intelligence research lab, using the Transformer neural network architecture, also known as the GPT-3.5 architecture, which is a model for processing sequential data with language understanding and text generation capabilities, in particular it will train the model by connecting to a large corpus of ChatGPT is not only a chatbot, but also capable of writing emails, video scripts, copywriting, translating, coding and other tasks.

On OpenAI's official website, ChatGPT is described as a language model optimized for conversation, and is the main model of the GPT-3.5 architecture.

ChatGPT has some of the features that similar products have, such as the conversational ability to answer contextually relevant follow-up questions within the same session. However, what has set the world alight in a short period of time is that in the screenshots posted by users, ChatGPT is not only able to converse fluently with users, but can even write poetry, compose and code.

ChatGPT is also trained to be ethically minded, saying "no" to unsolicited questions and requests according to a pre-designed ethical code. Valid answers will be denied if malicious intent, including but not limited to violence, discrimination, crime, etc., is found in text prompts given by users.

In late November 2022, ChatGPT, an artificially intelligent conversational chatbot, was launched and quickly became a social media sensation, with over one million registered users in just five days.

By the end of January 2023, ChatGPT had surpassed 100 million monthly active users, making it the fastest growing consumer app in history. 

On 2 February 2023, US-based artificial intelligence (AI) company OpenAI released ChatGPT Plus, a pilot subscription plan for ChatGPT. ChatGPT Plus will be available for $20 per month, giving subscribers access to a more stable and faster service than the free version, as well as the priority to try priority for new features and optimisations.

On 2 February 2023, Microsoft made an official announcement stating:All of its products will integrate ChatGPT across the board. In addition to the previously announced search engine Bing and Office, Microsoft will also integrate ChatGPT in its cloud computing platform Azure, whose OpenAI service will allow developers to access AI models.

February 3, 2023 - IT industry leaders fear that ChatGPT, the big-name artificial intelligence chatbot, has been used by hackers in planning cyber attacks.

A report by Black Berry, which surveyed 500 IT decision makers in the UK about the revolutionary technology ChatGPT, found that more than three quarters (76%) believe that foreign countries are already using ChatGPT in cyber wars against other countries. nearly half (48%) believe that in 2023 there will be malicious use of ChatGPT to cause "successful attacks". ChatGPT to cause a "successful" cyber attack.

On 2 February 2023 local time, the US artificial intelligence company OpenAI, the developer of ChatGPT, launched a paid subscription version of the app.

On 7 February 2023, Microsoft announced the launch of Bing (Bing), the latest version of its AI search engine powered by ChatGPT, and the Edge browser. "A new era for search engines", said Microsoft's CEO. At a press conference in Redmond, Washington in the early hours of the 8th, Microsoft announced the integration of OpenAI's long-rumoured GPT-4 model into the Bing and Edge browsers.

16 February 2023 - Microsoft's move to integrate AI chatbot functionality into its Bing search engine and Edge browser is beginning to show results, with 71% of testers approving of the AI-optimised Bing search results. 

On February 16, 2023, after conducting market research, the team at Paramount Intelligence Divebots decided to integrate ChatGPT, which is known for its "content generation and intelligent interaction", and officially launched the intelligent marketing assistant.

February 27, 2023 - Snapchat is launching a new version of its ChatGPT chatbot based on OpenAI. The bot, called "My AI", will be anchored to the chat tab in the app's interface, and while it will initially only be available to paid SnapchatPlus subscribers at $3.99 per month, the ultimate goal is to make the bot available to Snapchat's 750 million monthly active users.

On 15 March 2023, OpenAI officially launched GPT-4. GPT-4 is a large multimodal model, i.e., it supports image and text input as well as text output, has powerful image recognition capabilities, and has a text input limit raised to 25,000 words. GPT-4 is characterised by, firstly, its larger number of training sessions; secondly, its support for multiple forms of output input; and thirdly, its learning capability is even greater.

In March 2023, Google announced that ChatGPT - Bard was officially opened for testing.

On 24 March 2023, OpenAI announced that ChatGPT supports third-party plug-ins, lifting the restriction that it cannot be networked. 

In April 2023, ChatGPT has stopped paying for Plus, with OpenAI giving the reason that there is too much demand.


III. ChatGPT technical logic

Combined with ChatGPT's underlying technical logic, some media have listed potential industrialisation directions for ChatGPT in the short to medium term: inductive text-based jobs, code development-related jobs, image generation fields, and intelligent customer service jobs. 


IV. ChatGPT's core competencies

ChatGPT has received a lot of attention because of the introduction of the new technology RLHF (Reinforcement Learning with Human Feedback), which solves a core problem of generative models, namely how to align the output of AI models with human common sense, cognition, needs and values. ChatGPT is the result of technological advances in AIGC (AI- Generated Content). The model facilitates the use of AI for content creation and enhances the efficiency and richness of content production.


V. Limitations of ChatGPT technology

The upper limit of ChatGPT's capability is determined by the reward model, which requires a huge amount of corpus to fit the real world and requires a high workload and overall quality of the annotator. An ongoing process. If AI technology iterations are not as good as expected and NLP model optimisation is limited, the progress of related industry development will be affected. In addition, ChatGPT's profit model is still in the exploration stage, and the subsequent progress of commercialisation remains to be seen.


VI. ChatGPT Social Application and Impact

A survey shows that 89% of college students in the United States are using ChatGPT for their assignments as of January 2023. [4] 

In January 2023, Sciences Po announced that it had sent an email to all students and faculty requesting a ban on all AI-based tools such as ChatGPT, aimed at preventing academic fraud and plagiarism. 

ChatGPT can also be used in scenarios such as developing chatbots, as well as writing and debugging computer programs, and in literary and media-related fields, including creating music, TV shows, fairy tales, poetry and lyrics. In certain testing situations, ChatGPT outperforms even regular human testers in education, exams, and answering test questions.

Feb. 4, 2023 - Israeli President Isaac Herzog became the first world leader to publicly use ChatGPT when he delivered a speech written in part by artificial intelligence (AI) on Wednesday.

February 7, 2023 - ChatGPT is part of "a very exciting new era of tools" that can "allow teams and competitor teams to do really fun things more effectively," but the technology won't revolutionize the video game industry.

On February 8, 2023, Microsoft announced the launch of Bing, the latest version of its search engine powered by ChatGPT.

On March 11, 2023, Samsung's DS division began allowing employees to use ChatGPT, but in less than 20 days there were three leaks of confidential information, two related to semiconductor devices and one related to the content of a meeting.

As of February 2023, this next-generation conversational artificial intelligence then raked in 100 million users worldwide and managed to break the circle from the tech world to become the fastest-growing consumer app in history.


VII. ChatGPT use controversy

1、Bans and restrictions

Several academic journals have issued statements completely banning or severely restricting the use of ChatGPT and other artificial intelligence robots to write academic papers.

On March 31, 2023, local time, the Italian Personal Data Protection Authority announced that it would ban the use of ChatGPT, a chatbot, with immediate effect and restrict OpenAI, the company that developed this platform, from processing Italian user information. At the same time, the Personal Data Protection Authority opened an investigation. On the one hand, the Authority believes that on March 20 the ChatGPT platform experienced a loss of user conversation data and payment information for payment services. The platform, however, did not inform about the collection and processing of user information and lacked legal basis for collecting and storing personal information in large quantities. On the other hand, the agency also accused OpenAI of not using any filtering mechanism to verify the age of users, despite the fact that according to the terms of OpenAI it claims that the service faces users over 13 years old. Later that day, OpenAI said it had taken ChatGPT offline in Italy. 


2. Data breach

On April 4, 2023, a combination of media reports revealed that three confidential data leaks had occurred within 20 days since March 11, when Samsung Electronics in South Korea allowed some of its semiconductor business unit employees to use ChatGPT. Among the three cases of confidential information leakage, two of them are related to semiconductor equipment and the other one is related to internal meetings.


3、Official Response

Using ChatGPT to complete assignments and write papers is an unethical and unhealthy way to learn. Students need to learn to think on their own, understand knowledge and complete assignments on their own. Using ChatGPT to complete assignments can only help students get marks for a short period of time, but it cannot improve their learning ability and knowledge.

In response to the Italian Personal Data Protection Authority's temporary ban on the use of ChatGPT, OpenAI responded that it works "to reduce personal data when training AI systems like ChatGPT, because we want our AI to understand the world, not the individual." The company said it was willing to work closely with the Italian Personal Data Protection Authority. 


4. Defamation lawsuit

In April 2023, Reuters reported that Brian Hood, the mayor of Hepburn Shire, west of Melbourne, Australia, accused OpenAI's ChatGPT of defamation and may file a lawsuit against the company after the chatbot wrongly claimed he was guilty in a bribery scandal in response to a question. Once formally filed, this will be the world's first defamation lawsuit against generative AI.


5. Use with caution

On April 10, 2023, the China Payment Clearing Association said that recently, tools such as ChatGPT have attracted widespread attention from various parties, and some employees of enterprises have already used ChatGPT and other tools to carry out their work. However, such intelligent tools have been exposed to risks such as cross-border data leakage. In order to effectively address the risks, protect customer privacy, maintain data security and improve data security management in the payment clearing industry, and in accordance with the provisions of the Law of the People's Republic of China on Network Security, the Law of the People's Republic of China on Data Security and other laws, the China Payment Clearing Association has issued an initiative to the industry, advocating that payment industry practitioners use ChatGPT with caution.

On March 27, 2023, Sophia University of Japan posted a grading policy on its official website regarding ChatGPT and other AI chatbots. The policy states that the use of text, program source code, and calculation results generated by ChatGPT and other AI chatbots in any assignment is not allowed without the permission of the instructor. If these tools are found to be in use, severe measures will be taken.

On April 3, 2023, the University of Tokyo posted a document entitled "About Generative AI" on its internal website, stating that "reports must be created by the students themselves and cannot be created entirely with the help of artificial intelligence. 

At an entrance ceremony at Kyoto University on April 7, 2023, the university's president, Nagihiro Minato, said, "There are many problems with AI-generated papers, and writing them requires a lot of effort, but it strengthens your mind and your ability to think."

Tohoku University posted a notice on its official website about the use of generative AI and urged faculty members to rethink how they assign homework and how they conduct exams. For example, it suggests "checking how the AI will answer before assigning exercises and reports" and "switching to an exam format that requires students to write in class.


6. Launching an investigation

In April 2023, the French National Commission on Information and Freedoms (CNIL) received several complaints about ChatGPT, a chatbot owned by OpenAI, which it said violated the EU Regulation on the Protection of Personal Data (RGPD) by allegedly violating users' privacy and fabricating false information. The Commission is conducting an investigation into the complaints received in relation to ChatGPT.

On April 13, 2023, the European Data Protection Board (EDPB), the EU's central data regulator, said it is setting up a task force to help EU countries respond to the popular artificial intelligence chatbot ChatGPT, to facilitate cooperation among EU countries, and to exchange information on possible enforcement actions by data protection authorities.


Tags: chatgpt

Comments

Great Review