Author | Yuan Gungun
Listing | CSDN (ID: CSDNnews)
I don't know from which year, the 10-fold engineer has become an unattainable legend in the programmer world.
Those who make good use of various technical tools, have high programming efficiency, master multiple programming languages, and master new technologies at their fingertips are regarded as "10x Programmers" . Adhering to technical output on the blog and making open source contributions to popular projects seems to have 10 times more time than others.
In 2019, there was an unprecedented heated discussion on the topic of 10x engineers on Twitter, which triggered tens of thousands of comments from netizens. Most people yearn for 10x engineers and hope to learn from their experience and efficiency. It was also mentioned that 10x engineers are a competitive business environment, which alienates the profession of programmers and turns them into tool people who desperately pursue efficiency.
Regardless of whether 10x engineers are the standard for evaluating excellent engineers, improving development efficiency has always been a topic of concern to all developers. With the recent rapid advancement of AI technology, this wave of large models has also been swept into the development tools.
AI writes code for me,
Or am I writing code with the AI?
The emergence of programming language models and code generation tools has caught the attention of engineers, and many developers have tried them for the first time. The code generation tools at this stage are far more than automatic completion. They can not only write basic bubbling and sorting algorithms, but also write unit tests and code comments, which take up a lot of repetitive work that developers spend a lot of time on.
Whether it is a general-purpose large model such as ChatGPT or Bard, they all emphasize their own code generation capabilities during demonstrations. Not to mention the successive releases and active updates of products such as GitHub Copilot, aiXcoder, CodeGeeX, etc., have made the industry really see the hope of revolutionizing the development paradigm.
According to the statistics of the MAD (Machine Learning, AI, Data) ecological technology panorama updated every year by VentureBeat, at present, developer products serving AI infrastructure construction have blossomed, ranging from data science editors, data science platforms, integrated cloud services, and data generation . And labeling, machine learning resource operation and maintenance, AI hardware devices and edge computing, open and closed source large models , can support developers in all aspects of machine learning and artificial intelligence application development.
CSDN conducted a small-scale survey on developers' use of code generation tools, and almost all respondents have tried code generation tools. A considerable number of developers have integrated such tools into their daily development process and paid for such services.
Developers mainly use the method of installing plug-ins in the IDE, and some developers use locally deployed code generation services. At the same time, we have also learned that some developers pay to complete code generation in conversational windows such as ChatGPT Plus.
In the research, we clearly found that such code generation tools, intelligent programming assistants, and natural language code generation have brought changes to the development paradigm of developers. More and more experts, scholars, and engineers are deeply collaborating with AI code generation tools in their daily work, which greatly improves R&D production efficiency and expands personal programming capabilities.
In a recent survey by GitHub, two groups of developers were compared with a strict sample grouping based on the task of writing a web server program using JavaScript. Developers who use GitHub Copilot spend an average of one and a half hours faster than developers who purely complete their work, and they are also ahead of developers who purely hand code in terms of completion.
With the help of tools, can you become a 10x engineer?
If AI is regarded as its own competitor, the current code generation tools are far from being able to cope with all processes in the production process, and there are also compliance and security issues.
However, if code generation tools are regarded as collaborators and pair programming objects, helping developers find directions and methods, and learning from them, and inspiring each other, it will definitely be the most sensible choice at the moment.
In GitHub's survey, it was also found that for junior and intermediate level developers, the value of GitHub Copilot can be more obvious.
With rich AI infrastructure and various tools in the development process, ordinary programmers can also stack buffs on themselves and gradually develop into 10x engineers. We recommend that developers actively try various new applications and tools, absorb the experience of pioneers, and bravely innovate their own development paradigms.
Regarding the above issues, we will hold the NPCon 2023 AI Model Technology and Application Summit offline in Beijing on August 12 (Saturday) .
The event will focus on the theme of building an AI R&D base for the whole link, focusing on sharing in the aspects of computing power selection and scheduling optimization, model training and deployment, AI tool use and capability evolution in AI R&D, focusing on projects that developers care about Implementation process.
In the roundtable sharing session of "Let's Talk", we will discuss how future programming languages and development paradigms will affect our lives with the theme of "New R&D Paradigms and Developer Capabilities Evolution in the AI Era" .
How should developers and scientific researchers use tools to improve efficiency? How to learn AI-related knowledge and skills to enhance your competitiveness.
