0.4 seconds – that’s how long it took Google to find 537 million ChatGPT-related content on the Internet. During the same period of time, there were 607 million Bitcoin content found by Google. A controversial technology phenomenon for many years, it took Bitcoin 14 years of existence to produce more than 600 million search results. Meanwhile, a huge amount of content talking about ChatGPT appeared after exactly 62 days.
Mentioned ChatGPT – Ha Khanh (Cau Giay, Hanoi) can’t help but be excited when she shows off the chats with the computer that she has experienced. Since being lent her OpenAI account by a friend to try it out, ChatGPT has become her virtual friend for the past week.
As a student majoring in Literature at the University of Social Sciences and Humanities (Hanoi National University), Khanh often makes friends with the library. She could not have imagined that one day, the contents of the whole book could be summarized quickly after just one command line on her own phone. Although not yet perfect, ChatGPT still brings a completely different experience to other players chatbots different that Khanh ever experienced.
First appeared in November 2022, ChatGPT is a natural language processing program developed to create a tool capable of analyzing, processing and understanding human language.
This is the product of OpenAI, a company founded in 2015 by entrepreneurs and investors in the technology field. In it, Sam Altman – former President of Y Combinator (the sponsor of startups like Airbnb and Dropbox) is now CEO and plays an important role.
This startup is also sponsored by many powerful figures in Silicon Valley such as technology billionaire Elon Musk, LinkedIn co-founder – Peter Thiel. Satya Nadella’s Microsoft also made its initial investment in OpenAI in 2019.
Since its public launch, ChatGPT has quickly become a phenomenon thanks to its ability to read questions on a variety of topics and respond quickly.
Only 5 days after its launch, this computer program has reached 1 million users. This speed is far ahead of previous technology monuments such as Instagram (2.5 months), Spotify (5 months), Dropbox (7 months) or Facebook (10 months).
According to the statistics of Ark Invest, the number of ChatGPT users has surpassed 12 million people. This computer program appears more and more in people’s talks and discussions. In those stories, it is referred to as a new technology product that has the potential to change and reshape the world in the years to come.
Artificial intelligence push
Compared to current chatbots, ChatGPT brings more special emotions thanks to the softness in language usage and the ability to improvise conversations fluently.
ChatGPT can answer knowledge-related questions, create text snippets automatically on demand, translate text from one language to another… Thus, this program can be applications for many different purposes, especially in the role of a virtual assistant or automated call center systems for customer care.
In fact, although it has only been released for a short time, there have been countless user-generated usage scenarios for ChatGPT, from the role of an online dictionary, to challenging essay writing and editing skills. writing, replying to emails, even writing and scripting. Regardless of the role, it gives satisfactory results, at an acceptable level.
With the current “coverage”, ChatGPT can create a push, a trend of creating texts with AI in a very human-like way, similar to the trend of AI painting. very popular. In the not too distant future, communication with ChatGPT can even go a step further with the appearance of a virtual assistant with a vivid appearance and voice.
Since the appearance of ChatGPT, there have been concerns about AI’s ability to replace humans, especially in jobs that require the ability to synthesize, analyze, generate documents and respond.
Even so, ChatGPT sometimes writes down answers that sound reasonable, but are inaccurate or meaningless. The quality and accuracy of the answers generated by the system also depend greatly on the AI training process from the input data source.
The program can only support people in some aspects, especially those jobs that are synthetic, repetitive and cannot replace people. AI algorithms are still not able to produce high-quality content, especially in areas that require experience, deep expertise, and creativity.
Behind the glamor
The fact that the whole world was excited about ChatGPT was something that even the Open AI team itself could not have imagined. Open AI CEO – Sam Altman once confessed: “Demand with ChatGPT exceeded our expectations”.
The success of ChatGPT led to Microsoft’s decision to pour a rumored $10 billion investment in OpenAI to accelerate groundbreaking research in artificial intelligence. However, behind the flashy appearance of the ChatGPT craze, the downsides are not known to everyone.
As a result of an investigation conducted by Time, for ChatGPT to produce smooth, smooth conversations like now, it is a feat. The development unit had to find a way to remove false information, words that incite violence, hatred, sexism, sexual harassment, etc. in the process of training AI from collected content. collected on the Internet.
OpenAI hired a Kenyan software company, using cheap local labor to label the data. It is content taken from the darkest corners of the Internet, with information described as barbaric, beastly related to child sexual abuse, torture, murder and even suicide.
These workers are paid between 1.32-2 USD per hour. They had to read and label about 150-250 pieces of text each day over a nine-hour period. In exchange for the success of ChatGPT today, many of them have been emotionally traumatized and are regularly haunted by the countless gruesome content they have to label every day. Those harsh working conditions are the dark side of the AI picture.
The growth and popularity of ChatGPT brings many other challenges to the world. The most noticeable is its use to cheat in completing articles and theses.
Besides, there are many concerns that ChatGPT will be used to become a tool for scammers. Some hackers have begun to use the program’s capabilities for the purpose of scripting phishing attacks to spread malicious code.
Will ChatGPT change the world?
Before the phenomenon of ChatGPT being interested from many worlds, VietNamNet consulted a number of experts in the field:
Dr. Dang Minh Tuan – Director of CMC Technology Application Research Institute (CMC ATI)
The essence of how ChatGPT works is to answer questions based on machine learning, that is, learning on the training data and guessing the next word in a sentence according to the probability and weight of the association with other words in the sentence. question context. Therefore, ChatGPT is not a knowledge database and not a semantically understandable system.
The advantage of ChatGPT is that it gives very similar answers to real people and because it is trained on a very large data set, it can answer quite correctly in many fields. However, it is not capable of creating new knowledge.
In a way, ChatGPT can be considered as an information search engine (search engine) with rich contextual request input and an answer for each request.
Unlike Google – the input is keywords but the output includes many results are links, ChatGPT gives more concise results, but has the weakness of lack of evidence and citation. Therefore, the results returned by the program are uncertain about accuracy and reliability. Over the past few days, there have been many examples of ChatGPT giving false results in both information and logic.
ChatGPT cannot replace humans in the near future. This can be a good reference tool, quickly giving answers in many areas. However, using these answers requires verification, which sometimes takes more time to find verified information than using Google.
The appearance of ChatGPT is similar to Google Search nearly 30 years ago. At that time, people were also worried that all knowledge is available on the internet, only Google can have it, so the profession of journalism or teaching will gradually disappear. In fact, Google is still just an information search engine and cannot remove some professions.
Tran Viet Hung – Founder and Chairman of Got It
We need to let ChatGPT run for a while longer, carefully observe and evaluate before drawing conclusions.
People often underestimate AI in terms of writing as well as language comprehension. But at the current rate of development, the next versions of ChatGPT will do even more miracles than now.
Although things have just begun, ChatGPT has contributed to increasing productivity for many people. This tool is especially effective when applied to writing-related tasks or information questions about general knowledge.
Instead of challenging ChatGPT like a puzzle, we should change the perspective to see what AI can do and turn it into a tool. This is what 99% of people around the world need.
Ngo Minh Hieu – Security expert
OpenAI uses pre-existing data sets collected from multiple sources to teach ChatGPT. With the current level, it can be considered as a source of reference information, combined with further verification by other sources. Users should not rely too much on ChatGPT, because the accuracy of the answers sometimes matters.
Professional programmers will not be replaced by this computer program. Instead, ChatGPT can be used to help developers in suggesting faulty code, solutions, adding features to improve the product.
If provided with a good input data source, ChatGPT will become more and more intelligent and be used more widely, even in fields such as art, music, literature….
Technology has always had a light and a dark side. On the other hand, if not “taught” well, without data screening, ChatGPT will become a disaster. Hackers can take advantage of this program to write fraudulent email content, fake news, suggest creating malicious code.
Some amateur hackers have been using ChatGPT for malicious purposes. This is an issue that security researchers are concerned about.
With the rapid development of AI, there will be a series of new forms of fraud such as using technology to impersonate a human voice or disguise real identity for the purpose of deceiving, sabotaging or smearing others, even even bypassing security systems. This is something that has happened with the appearance of DeepFake before.