Use ChatGPT to get rid of junk Internet content!

ChatGPT is a chatbot that knows astronomy from above and geography from below. Screenshots of conversations are spread all over the Internet - life is incomplete without chatting with it about the Qin Emperor, Han Wu, and the new coronavirus.

Just as Asian teams lost one after another and bid farewell to the 2022 Qatar World Cup, ChatGPT gave an exciting answer. Maybe, in a parallel world, we had the glory of being selected for the World Cup twice.

In the past few days, everyone has watched too many amazing performances of ChatGPT. Whether it is writing a paper or writing code, it can be done in one go and with ease. But its answers still have certain limitations and are not without errors. If you know the correct answer, feel free to browse. If you don’t know, it’s better to be cautious.

The far-reaching significance of ChatGPT is that it raises the lower limit of all articles in the world.

The vast majority of articles flooding various channels are far inferior to ChatGPT in terms of accuracy, knowledge and logic. "As we all know", there are a lot of manpower on the Internet, either making headlines, producing hydrology, or concocting vulgar topics. Gary Illyes, a trend analyst at Google Webmaster, said that 60% of the content on the Internet is duplicated.

However, no matter how good the article written by AI is, it is only a "short and fast" data integration.

Writers should not be satisfied with the routine of cyber "shadow writers" like ChatGPT; readers should pursue more in-depth and higher-quality reports.

Santa Claus and Christmas Tree

ChatGPT is particularly good at dealing with some life problems, such as family disputes, emotional problems, etc.

Last week, a screenshot of ChatGPT was circulated to help people answer the question "Wife is angry that her husband thinks the Christmas tree is too expensive and won't buy it." ChatGPT first tries to break up the argument, saying that it's something to discuss and don't get emotional, and then gives a solution. ——Either the two of them can make a Christmas tree by themselves at no cost; or they can go on a trip so that they don’t have to worry about the quality of the Christmas tree at home.

This routine is completely in line with psychological theory:

calm down the emotions, stand on the side of the speaker, and then rationally provide some practical solutions. From this perspective, ChatGPT is like the most caring and lovely friend.

"Christmas" is a high-frequency question in December. Shopify executive Cynthia Savard “broke defenses” on ChatGPT. ChatGPT helped her explain Santa Claus to her son: "I am writing to let you know that I am not a real person, but a character in the stories your parents told you out of love."

"Your parents brought joy and magic to your childhood by telling you stories about me and my elves. They wanted you to believe in the spirit and the magic of the holidays."

The article writes emotionally, "The love and care your parents have for you is real. They have created wonderful memories and traditions for you, and they hope your childhood will be special."

Two "Christmas" stories actually

highlight the real ability of ChatGPT as a new generation of "chat generation pre-training converter": to answer questions raised by humans as comprehensively, appropriately and informatively as possible .

Answer according to the human mind

ChatGPT comes from the OpenAI Research Laboratory and is powered by the GPT-3.5 series of models, including 3.5 Previous model versions were trained using text and code data on the Azure AI supercomputing infrastructure.

It should be noted that the preview version currently used by people is not connected to the Internet, and any responses from it come from the offline training model. You ask it "What's the weather like today?" and it doesn't know.

The most important change in the GPT-3.5 series models is the adjustment based on real human feedback. This is a newly used AI training method. The tagger will write the expected reply in the model, sort the tagged answers according to the expected reply, and reward the model through ranking. In the process of continuous iteration, the reward model is input and the optimization parameters are obtained.

ChatGPT training steps

For example, it’s like you are taking a final exam and fill in the exam paper according to the content you have studied throughout the semester. If you answer correctly, Get high scores, rank high, and go home with the latest Super Mario Lego your parents bought. If it’s a Chinese language exam, you know that the marking teacher likes Lu Xun. When writing the essay, he never forgets to say, “In my back garden, you can see two trees outside the wall, one is a jujube tree, and the other is also a jujube tree.” classes, and the chance to unlock more rewards.

After adding a large amount of human-expected mark feedback, it is equivalent to a "tendency guidance". If you speak according to the same mechanism, you will be rewarded. As a result, ChatGPT has greatly surpassed its predecessor GPT3, which could only dryly answer the "definition" of keywords based on questions, while ChatGPT can already speak "according to human meaning".

Machine learning and human brain learning are generally the same. This is also the reason why artificial intelligence suddenly made rapid progress five or six years ago: the deep learning model convolutional neural network, a multi-layer perception system similar to artificial neural networks, achieved a breakthrough.

Different neural layers receive and process different information, and they progress layer by layer to obtain a "correct" description of something.

The computer simulates this process and begins to "learn" people's "recognition" and "cognition" abilities.

# "Reward" is similar. In most cases, any activity that humans do that is beneficial to survival will secrete some reward substances such as serotonin, oxytocin, endorphins and endocannabinoids in the body, making the brain feel "happy".

Getting a high score determined by the system is a "reward" for AI.

Writing is a kind of dignity

ChatGPT has fully possessed all the qualities of excellent customer service. Compared with the stupid customer service robots that people most often encounter, ChatGPT can definitely make all consumers happy and regain the feeling of God who no longer has high blood pressure.

All industries that require customer service are likely to benefit from the advancement of this technology.

In addition, ChatGPT is also a good efficiency tool.

For example, search. Some people think that it is enough to replace search engines such as Google, but it may be too absolute. First of all, ChatGPT's integration of data is based on "dialogue", which adds a lot of nonsense that will waste searchers' time. Secondly, ChatGPT does not absolutely guarantee authenticity, nor does it provide sources of information, so searchers cannot even verify it from multiple aspects. Under the premise of fidelity, some industries that require knowledge search and inspection are expected to use ChatGPT.

In other words, GPT-4, which has a larger database and is connected to the Internet, will be a big killer that can compete with search engine giants.

For example, writing. Not just ChatGPT, GPT-3 is already excellent at writing papers. They will arrange the information they collect under the very typical TOEFL essay writing ideas, which is very formal and standardized, but there is really no "original idea".

When it comes to writing code, ChatGPT is a blessing. If you install a ChatGPT plug-in on your browser and ask for code questions at any time, you are likely to avoid many detours on the road to Alibaba P8.

Jonas Degrave shows how to turn ChatGPT into a Linux terminal and interact with the virtual machine directly from the web browser (manual dog head)

For example painting. ChatGPT can be linked with the recently popular Stable Diffusion. Use ChatGPT to form a piece of text that meets your requirements, and then input the text into StableDiffusion. The resulting paintings are generally much better looking than directly entering the description yourself. Maybe AI understands AI better - ChatGPT's description is more detailed and easier to extract.

As a professional writer,

ChatGPT and other AIs that can "impersonate" human writing are like cyber "shadow writers", promoting people to reflect on the meaning of creation.

In a sense,

ChatGPT is another aspect of Q&A websites and payment for knowledge. If you want to know a skill or a piece of information quickly, ChatGPT will answer you immediately and without hesitation. Reading and knowledge can be acquired through very convenient and simple technology, rather than through a long and tortuous learning process. Unlike search engines, you don’t even need to filter information and identify the source yourself, ChatGPT does it for you.

For creators, this is a big challenge. The goal is to be fast and direct. Watch a movie in five minutes and explain a classic in ten minutes.

ChatGPT is a technology that meets the needs of the times, but it should not be the purpose of creation.

Articles created by cyber "shadow writers" will soon flood various media and channels, and many people will be unable to tell the difference between them and human writers. Previously, "Space Opera" painted by Stable Diffusion won an award, but at first the judges didn't know it was painted by AI.

This is a normal phenomenon. In other words, AI’s high-quality texts are enough to be circulated and commendable.

But if human creators were satisfied with this, there might never be another "Macbeth".

ChatGPT, as well as the previously released Notion.ai, can help creators quickly generate drafts or outlines, saving time and energy on the details and depth of writing, creating more valuable creations s work.

With AI, it would be unreasonable for us not to create great works earlier.

This article comes from the WeChat public account: Ai Faner (ID: ifanr), author: Rong Zhihui