ChatGPT is another new gift given to mankind by the technology geek Musk. When Musk founded OpenAI in 2015, he may not have realized that such a great product would be born under its umbrella. This is a phenomenal application, with more than 100 million daily active users within two months of its launch. When ChatGPT appeared in the public eye, I was not too keen on it. I thought that the combination of CHAT GPT was just a carnival of capital. Before ChatGPT became popular, GPT 3.0 was actually popular for a while, but it was not as popular as ChatGPT. Of course, the GPT version of ChatGPT has been upgraded to 3.5, which is more powerful than 3.0. GPT is a pre-trained model based on the Google Translate Transformer algorithm model. Behind it is the huge computing power and logical computing power of CPU and GPU. Coupled with massive data and a large amount of manual training, ChatGPT has more than 170 billion parameters.
A while ago, various jokes about using ChatGPT to replace human work have been circulating on the Internet. It makes people think that ChatGPT is a force that subverts human society. Is this really true? Only by trying it yourself can you have the best say. However, there are still certain thresholds for using ChatGPT in China. First, you must be able to access the Internet scientifically, otherwise you will not be able to access OPENAI’s website in China. Second, you must register an account with a foreign mobile phone number. OPENAI is currently subject to export restrictions by the U.S. Department of Commerce. , membership registration and services are not open to China.
I asked a question about learning PG, hoping that it would give some suggestions. The answer was quite surprising, given my understanding of semantics and my understanding of this question. The answers are all quite satisfactory, and basically reflect the preliminary steps of learning the PG database. They do not answer randomly and mislead students.
I will continue to ask and it will give me a more detailed explanation. From this point of view, ChatGPT’s semantic understanding and multi-round dialogue, as well as text generation capabilities All are extremely powerful. You can see that I gave a feedback in the upper right corner and approved the answer. In fact, I am also helping ChatGPT train data and improve the capabilities of the model. The more audiences use ChatGPT, the more powerful its model will be.
Next I asked a question, what should I do if WAL is broken? The answer was quite satisfactory. But the answer to the next question is a bit retarded.
I saw that it did not use pg_resetwal when answering, but used pg_resetxlog, so I teased it a bit, but unexpectedly I immediately discovered its misunderstanding. In fact, the pg_resetxlog tool introduced in PG 9.6 was renamed pg_resetwal after PG 10. It should be the same tool.
Starting from this mistake, ChatGPT’s answer was a mess. This is similar to humans. It revolves around a wrong cognition, even if the logic is correct, the cognition is generally The above are all wrong.
In fact, during the use of ChatGPT in the past few days, when faced with some complex problems or some exploratory academic discussions, ChatGPT often Unsatisfactory. Most of the papers it recommends are not of high quality and have no reference value, and several of them even do not exist.
And then, it also had no idea what to ask. When I asked for the DOI number of a certain article, it told me that his training set ended in 2021, and suggested that I search elsewhere. During the time I used it, I felt the powerful capabilities of ChatGPT and also learned about the limitations of ChatGPT. Most of what is said on the Internet about writing programs and papers for others is not very reliable. I don’t rule out that ChatGPT can help people write some specific framework codes, but some simple programs and low-code tools oriented to industry business characteristics are far more efficient. Writing programs through chat is more efficient than serious working programmers. Efficient, that programmer's level has been improved many times.
ChatGPT is extremely efficient in replacing traditional search tools to find information. I think this application will pose a huge challenge to Google search. Through multiple rounds of dialogue, we can fully understand the needs of searchers and then provide search results. This can It is a very effective method to obtain more accurate information from massive information and save search costs. Back to the topic of DBA, ChatGPT is a good helper for DBA, and it is absolutely very efficient for searching technical information. However, ChatGPT's answers are not always correct. I once had a conversation and asked him what to do if the current redo log file is damaged. The answer was full of loopholes, and the script given was not very reliable, so when learning, It is very important to identify errors. At least ChatGPT is still far behind the level of an Oracle database expert, so DBAs don't have to worry about it taking away your jobs.
ChatGPT is amazing, and for the current CHAT capabilities, ChatGPT represents the highest level of mankind. But as I said at the beginning, ChatGPT is not real intelligence, it is the output of algorithm, capital, computing power, and data training. It is not necessarily higher in technical level than some other AI products, but ChatGPT opens another door for us, that is, high-level AI models can also be used out of the box. OPENAI has opened the door for humans to explore artificial intelligence. I think a large number of companies will soon follow this track, and more applications will be developed. This is ChatGPT’s greatest contribution to the development of science and technology.
Recently, some friends in the surrounding area are discussing how to use ChatGPT for some commercial applications. This idea is still relatively difficult to realize in China. It requires scientific Internet access and avoiding OPENAI’s blockade of users in China. These two major Risk determines the infeasibility of directly using the OPENAI platform to develop commercial applications. In addition, OPENAI will soon launch the commercial version of ChatGPT PLUS. It is said that the monthly fee is 20 US dollars. I think once the commercial account is launched, it will be a certainty to set limits on free accounts. It is not that easy to make money. A few days ago, I encountered a situation where there were too many questions within an hour and I was asked to try again after an hour.
However, domestic companies are also ready to move on this track. I think domestic ChatGPT will also appear within a year, but I don’t know how different its level is from the genuine ChatGPT. If it is like the gap between Baidu and Google , which is just about usable.
When discussing with me, he also asked, why can’t such great IT products appear in China? Why can only applications like Taobao and WeChat be produced? Is it really that my skills are inferior to others, or that my thinking mode is backward? In fact, the most important thing is not these problems, but the difference in the attitude of capital. The patience of the capital behind a company like OPENAI is very strong, and it can support an unprofitable company for ten, twenty or even fifty years. But it is a project that has great potential for future technology. Domestic capital, on the other hand, will be impatient if there is no substantial return within five years. For a project like ChatGPT, even a small goal cannot accomplish anything. Even if someone wants to do this, will they have the capital in China to dare to invest? Even if capital dares to be invested, which young people with technology in their arms dare to ask for 3.5 billion US dollars to do this? I admire Musk. What I admire most about Musk is that he dares to brag. If he doesn’t even brag, what goal can he seek to achieve? Turning magic into decay and turning decay into magic are always two necessary steps in creation. As I write this, thinking about Musk, I want to watch the drama "Silicon Valley" again.
The above is the detailed content of Can ChatGPT make DBAs unemployed?. For more information, please follow other related articles on the PHP Chinese website!