GPT-4 has only been born for a few days, and I feel that the circle of friends has been swamped by this artificial intelligence. While everyone is lamenting the vigorous development of the artificial intelligence industry, many people are suffering from AI anxiety.
Among them, programmers bear the brunt. There is no other reason, only because of the amazing ability of GPT-4 and the speeches of many big shots.
GPT-4 can instantly create a simple website with only an extremely scribbled hand-drawn design. It is considered by many to be a "front-end engineer's nightmare".
Later, designer Pietro Schirano used GPT-4 to rewrite the classic game "Pong" and successfully completed the task in 60 seconds.
Matt Welsh, a professor of computer science at Harvard University and head of senior engineering at Google, spoke at the Association for Computing Machinery (ACM) conference in February this year, and also believes that the emergence of ChatGPT heralds the beginning of the end of the career of programmers .
Everyone seems to be bad-mouthing programmers. But in this way, can it replace human labor? Borrowing the complaints of netizens that is:
"Jianzhan came out, and you said that programmers would lose their jobs.
When low code comes out, you say that programmers will lose their jobs.
Copilot came out, and you said that programmers would lose their jobs.
Chatgpt came out, and you guys said programmers would lose their jobs..."
I don’t know how many times it has been said that programmers will lose their jobs, but the “code farmers” are still strong.
Laymen will only think that programmers can only write codes. Only when they have been programmers can they realize that the time spent writing codes is at most about 20%~30%. The rest of the time is spent debugging and modifying codes. There are also meetings to discuss needs and so on.
Therefore, even if ChatGPT helps programmers save all the time of writing code, it only saves 20% of the time.
Assuming that ChatGPT is asked to write the first-generation version code, when functions are added later, it is still necessary to manually read the code first, and then proceed to the next step of writing after reading it.
Zhihu has a netizen who left a message: When some self-media started to brag about ChatGPT, I once thought that it could really write code for me, but it turned out that it was impossible.
There are two main reasons:
1. The degree of intelligence is not enough, and complex codes are really impossible to write.
2. I didn't communicate enough with him, because my code is based on my business logic. Although ChatGPT is very powerful, but when it comes to my own business, I really can't teach it with a few sentences, so it can't understand what I want do what.
In addition, there are still many other problems in GPT-4.0. For example, it could fall into similar problems as its predecessors, such as generating misinformation or wrong answers.
In addition, the energy consumption of GPT-4 is also a problem due to its large size and complex training model . Although GPT-4 has many excellent skills, it still cannot replace human creativity and thinking ability, but is only a tool and assistance.
Overall, the release of GPT-4 marks another leap forward in artificial intelligence technology, which will further promote the development of the IT field and bring us more convenience and opportunities.
Just like the Internet ten years ago, what we have to do is to make good use of this change , and machines can be upgraded and iterated, so can humans. The source of our own self-improvement is continuous learning. Only in this way can we be less afraid of the impact of ChatGPT.