ChatGPT was released on November 30 last year, only 2 months after its listing, the monthly active users have exceeded 100 million, and it has been pushed on the hot search many times in February. If the first wave of ChatGPT has not yet attracted too much attention, the release of ChatGPT-4 directly dropped a bombshell that may affect 80% of the work. Musk thought it was scary, even its founders said they were a little afraid of it, and Bill Gates gave it a high opinion of "one of the most revolutionary technologies."
So what exactly is ChatGPT? Why does it frighten many tech industry bigwigs, but give it a high rating? Are they even competing to imitate, learn, and study? What can it do? Who are the workers affected? This article talks to you about these ChatGPT issues that everyone is more concerned about, as well as my ChatGPT experience as an editor.
What is ChatGPT?
In the most popular terms, ChatGPT is an artificial intelligence that can chat. For a formal explanation, ChatGPT defines itself as "a large-scale language model developed by OpenAI that understands and generates natural language that can answer questions, provide suggestions, and have conversations."
In order to facilitate understanding, you may wish to think of ChatGPT as a "brain", in order to make it smart, we must continue to train it, just like humans, through learning to improve their own knowledge! As you learn more and more knowledge, it will also become smarter, and when all the knowledge in the world and the experience of all people are added to it, this will become a very terrifying thing!
When there is enough knowledge learned and the variety of knowledge learned is rich enough, ChatGPT will become an omniscient "immortal", knowing astronomy and geography, and ancient and modern.
The key is that ChatGPT is not simply a summary of information, similar to a search engine, through the search keywords to give a direct answer, but when receiving this question, through the judgment of ChatGPT, give a relatively accurate answer!
At the same time, ChatGPT has the ability to draw inferences from one to another, can summarize the rules, for some things or problems that have not appeared, will get the answer by drawing inferences, maybe this answer is wrong, but when the knowledge is learned enough, the learning and judgment mechanism continues to upgrade, the closer the answer evolved from this calculation is to the correct.
What jobs might ChatGPT replace instead?
According to the OpenAI study, GPT is GPT: An Early Look at the Potential Impact of Large Language Models on the Labor Market, about 89 percent of the U.S. workforce is likely to have at least 10 percent of the job impact as a result of GPT. Among them, customer service personnel, translators, writers, journalists, editors, marketers, accountants, human resources and other industries will be greatly affected.
Because ChatGPT is an artificial intelligence built by knowledge and experience, the first wave of affected is likely to be "brain" workers, on the contrary, the impact on physical workers, the impact may not be so great, but not without impact, when we put ChatGPT this "brain" "installed" on the robot, then perhaps ChatGPT can complete more work.
Code capability
Some time ago, the latest ChatGPT version 4.0 was just released, and one of the biggest improvements was the ability to support image input and, of course, the accuracy of knowledge brought about by the increase in the database. OpenAI officially demonstrated the functions that ChatGPT 4.0 can achieve, and the most exciting thing for me is that you only need a hand-drawn sketch to generate the corresponding web code, which is also closely related to code farmers.
Unfortunately, even if we spent $20 and qualified for ChatGPT 4.0, we still couldn't experience the image input function.
However, I tried to manually enter it to write a front-end page code that imitates Taobao's html, and ChatGPT does give a string of code, and gives the corresponding text explanation and a tutorial for adding more products and footers.
Since I only know the code halfway, I can't give a more accurate judgment, but from the perspective of the implemented functions, it may indeed be helpful for code farmers. But if you say that using ChatGPT to completely replace the code farmers, it is obviously not possible at present, but you can use this framework to optimize in detail.
Editing capabilities
As a media practitioner, it is the industry where ChatGPT is greatly affected, and since I can't experience image input at present, I focused on experiencing the text input function of ChatGPT 4. The three categories of press release (real-time), historical article (past) and evaluation article (thinking) were mainly tested, but it should be noted that the training database of ChatGPT 4.0 is as of September 2021.
The press release mainly tested two manuscripts, 1 is the release press release of the 2020 iPhone 12 series, and 1 is the release press release of the 2022 iPhone 14 series, the former only a small part of the content error "For example, the chip of the iPhone 12 is A14, and it is written as A15, and the pre-installed system is written as the iOS15 system", and the iPhone 14 has more errors, such as only writing 3 models, the processor uses A18, it is simply in " a serious book of nonsense", and some clichés.
As we said above, when ChatGPT encounters knowledge content that it does not know, it will make up through its own "inference" logical ability, and the fabricated content will be wrong, such as the content of the iPhone 14 series above, but the general idea and logic of the article are not obviously confused.
When writing the article "The History of Huawei's Smartphone Development", ChatGPT4 retrieved the data in the database and integrated these data to produce a more reasonable article, the paragraphs are orderly and clear, but if you read carefully, you will feel that the article is still familiar with more clichés, and there are still some errors, and these need to be judged. But it provides development history nodes and historical content that can be used as a reference when writing articles.
When writing iPhone 12 Pro review articles, the content of the article is basically the stacking of parameters and some clichés, these review articles are difficult to have reference significance, and it is difficult to make people read patiently, in the final analysis, the article lacks the "humanistic" atmosphere, lacks the thinking and views of the product.
From the actual experience, at least for now, it will take some time for ChatGPT to replace the ability of journalists or editors. Even if ChatGPT has the ability to access information in real time in the future, so that it can create content in real time for what is happening at the moment, but when creating, it can only rely on the learned templates and logic to create, and lack of human thoughts, human emotions, which is the biggest shortcoming of ChatGPT in text creation.
However, ChatGPT can become a good helper, through its information collation, analysis, we can obtain information more quickly, only need to compare the accuracy of the information, and the text content to modify, moisturize, which will obviously improve the efficiency of our creation.
Literary creation
Similar to editing, literary creation is also the generation of written content, and I try to write a story outline and then use ChatGPT to complete the story. ChatGPT4 was soon completed, and I also introduced new characters to my story content and expanded the story line, such as playing chess with my brother and knowing a person who met the one-eyed dragon.
And if we want to control the direction of the story, we can continue to talk to ChatGPT, to guide ChatGPT, let it continue to compile the direction of the story, and may only need us to modify the details to successfully write a novel. If we guide in different directions, the content of the story will naturally be different, so ChatGPT is just a tool that can help literary creators create content, and the core story content still needs to be controlled by us.
Other capabilities
In addition, I also experienced ChatGPT's translation, text generation tables, mathematical calculus, classical Chinese translation, marketing and other capabilities, relying on ChatGPT to learn a lot of knowledge, you can get relatively good results, and will help in work and study.
Translation Test:
Text to table:
Classical Chinese Analysis:
Math Answer:
Marketing Tagline:
What industries does ChatGPT affect?
So what industries does ChatGPT affect? When asked ChatGPT, it also gave its own answers, including customer service, education, content creation and editing, translation and language learning, human resources, market research and data analysis, healthcare, financial services, smart home and IoT, gaming and other industries, and answered the impact on each industry.
In my opinion, it's not just ChatGPT that has affected the industry, it's essentially AI that has affected all walks of life, and now AI is already influencing. For example, customer service has now provided AI artificial customer service; In the education industry, AI products have been released, such as AI dictionary pens, and smart homes and the Internet of Things have long integrated AI functions; The game industry has also joined AI technology to make the game picture quality more realistic, etc., and there are now many such cases, and there are many industries that are being changed by AI, and AI is slowly affecting the world.
Finally, I want to say: don't be afraid of ChatGPT, but be aware of learning
Finally, we don't need to be afraid of ChatGPT, it is just a tool extended by technology, just like the stone tools of the Stone Age, the metal of the Bronze Age, the steam engine of the steam age, the Internet of the computer age, this is the progress of science and technology, but also the progress of the times, the birth of these tools is to help us, not to ban us, what we have to do is to learn to use them, use them to make science and technology develop further.
Then again, don't worry or be too anxious! The work behind ChatGPT also requires a lot of capital investment, and it is also necessary to pay the corresponding fees when using it, the key to using technology similar to ChatGPT, how much profit can it bring? When you can't make ends meet, it's clear that manpower is more cost-effective. However, everyone should have a sense of crisis, how can they not be eliminated in the next era? How can I follow in ChatGPT's footsteps? It is a question that everyone has to think about, and keeping learning should be the right choice!