The crystallization of ChatGPT

Tram Ho

Many organizations and individuals race to develop tools to tame ChatGPT and the like

Right from the early days of 2023, ChatGPT super chatbot has quickly caused technology storms on a global scale. Indeed, what it has done now has made people both happy and anxious.

Soon to build legal corridors

In December 2022, in The Wall Street Journal (WSJ-USA), Oren Etzioni, CEO of the Allen Institute for AI – a non-profit organization specializing in AI research and engineering, said: “Six months from now Now you’ll see amazing things you’ve never seen before.” He made such a prediction about AI applications only about half a month after the super AI application ChatGPT was officially released by OpenAI Technology Company on November 30, 2022.

Just 5 days after its release, ChatGPT – although limited to only a few countries, already has 1 million users. According to Genevieve Rock-Decter/CFA, before that, to get the first 1 million users, the fastest Instagram also took 2.5 months; Facebook took 10 months, Twitter took 24 months, Netflix took 41 months. According to Bret Winton, Head of Futures at ARK Venture Investment, ChatGPT has surpassed 10 million daily users just 40 days after its official launch. (Instagram took 355 days to reach 10 million registered users.) By January 31, ChatGPT had reached 100 million users and OpenAI on February 2 began to release a paid version of ChatGPT Plus that costs 20 USD/month with more super features than the free version. Thus, with the case of ChatGPT, from now on, the world must witness AI reaching door to door and in the hands of everyone widely. Previously excited to receive but also had to be wary of AI, now we have to upgrade the warning even higher. In fact, in the past time, the technology giants developing AI have had to develop ethical rules on AI applications – even avoiding AI research and development in some fields for some purposes.

Khắc tinh của ChatGPT - Ảnh 1.

The South China Morning Post reported on January 30 that China’s “Internet search giant” Baidu plans to launch a ChatGPT-style chatbot in March.

Professor Mindy McAdams of the University of Florida School of Journalism (UF, USA), a JavaScript and Python programmer, in an article on AI in media and society in July 2022 said: “Perhaps The biggest misconception about AI is that machines will soon become as smart as humans or even smarter than all of us.As a common feature in science fiction books and movies, the idea of ​​a a super-intelligent computer or robot that occupies a strong place in our minds but not in the real world.No AI system that achieves impressive results is truly intelligent the way humans are. intelligent people (even babies!)”. On the technology page of the WSJ, the two authors Hao and Kruppa wrote a long article in July 2022 titled “Tech giants pour billions of dollars into AI, but the hype doesn’t always fit the bill.” in line with reality” analyzes the investment of billions of billions in AI by technology corporations Google, Meta and startup OpenAI. “Researchers have noted that the belief that AI is becoming – or may become – conscious remains marginalized in the broader scientific community,” they said.

AI technology simulates the human brain, but it is more powerful than the human brain by being able to search, absorb and process vast data and as fast as the blink of an eye. One thing to note is that AI is not only constantly developing and upgrading, but can also develop itself through higher levels such as Machine Learning, Deep Learning… If in the past, people mainly focused on development. Developing and applying AI, now, after ChatGPT was born, experts have to worry about researching ways to control, tame and limit the harms when AI is abused by people for bad purposes. even crime. Other technology companies, of course, have the added incentive of being competitive, seeking to hold back their rivals’ expansion.

Authorities must certainly have appropriate responses to be ready to deal with illegal acts of abusing and abusing AI tools. Authorities must also soon build legal corridors to maximize the benefits of AI, while minimizing its dark sides.

Using AI to control ChatGPT

In terms of technology, experts use the “let alone poison” technique, using AI to treat AI. Here ChatGPT is just one example. Of course, what one can do right away is find a way to detect content “composed” by ChatGPT. Even if this can be done, it is only limited to documents and works such as essays, articles, poetry … written by ChatGPT at the request of users. No one can predict what “scientist” ChatGPT can do at the request of users in private chat corners. Despite being super intelligent, it is still a technological tool created and trained by humans. The “father” of ChatGPT, OpenAI, has confirmed that they are working on a tool to help detect content generated by ChatGPT.

Currently, a “crush” of ChatGPT is the DetectGPT tool developed by a group of researchers at Stanford University (USA). True to the name “Detect GPT”, DetectGPT has the function to detect texts generated by ChatGPT. This method uses the very features of the Large Language Model (LLM) engines on which ChatGPT is based to evaluate whether a text is generated by the LLM engine or not. According to the researchers, DetectGPT is now 95% accurate. With tools like DetectGPT, if perfected, editors, teachers, etc. can detect whether articles and essays are “required” by AI tools.

While hundreds of millions of mostly curious and fun-loving people around the world are eagerly signing up to use ChatGPT, on the other hand, more and more organizations and individuals are racing to develop tools to tame ChatGPT. and similar tools. One of them is Edward Tian, ​​22 years old, a final year student of computer science and journalism at Princeton University (USA). With what was gathered through studying how to detect text written by AI systems while working at the Natural Language Processing Laboratory of Princeton University, Tian wrote the GPTZero application in just 3 day at a coffee shop in Toronto (Canada) during his return home for the Lunar New Year of the Rabbit (Tian is of Chinese descent).

Basically, GPTZero uses ChatGPT against ChatGPT itself. This software will check if the text is “no or too much involvement” of the AI ​​system. After launching on January 3, the GPTZero application has surpassed the milestone of 80,000 visitors. Tian said he will make the app free for anyone who needs to be able to use it.

ChatGPT graduation thesis writing scandal at Russian university

The Russian National University of the Humanities (RGGU) must call for restrictions on access to ChatGPT. The reason is because a student of the school has successfully defended his graduation thesis that was completely written by OpenAI’s artificial intelligence (AI).

The scandal broke out at RGGU after a new graduate of the school took to Twitter to share “all experiences” using ChatGPT to write graduation thesis. Although the thesis was completely created by the AI ​​of the US-based company, this student still successfully defended it. “This student took only 23 hours to complete the graduation thesis compared to the few weeks that other students in the same course had to spend,” – according to RT station. The male student’s share above has received mixed reactions from the public. Some praised the man as “good”, while others accused him of cheating. Others were so angered by this tactic that they wrote complaints to the RGGU and the Russian Ministry of Education, calling on officials to investigate and cancel the thesis.

Before the reaction from the public, RGGU also quickly condemned this student, and called on the authorities to restrict access to ChatGPT for educational institutions. “Decades ago, the main problem for universities was plagiarism. However, this problem has now been successfully solved. Now, the community is facing a new challenge related to plagiarism. using AI in scientific and educational activities. Russian and foreign experts have expressed concern about this problem, and finding solutions to prevent it is the top task,” – RGGU’s proposal has a paragraph. . Regarding the incident, the RGGU Board of Directors announced that they would re-evaluate the aforementioned controversial thesis with the participation of school teachers and outside experts.

In response to the news, Deputy Head of the Scientific and Higher Education Committee of the Russian State Duma, Aleksandr Mazhuga, urged RGGU not to cancel the graduation thesis because it had been approved before the university’s scientific council.

Share the news now

Source : Genk