You must have heard a saying: You should start learning computers from childhood. In the past many years, learning to program has been a fashionable thing, but with the development of generative AI technology, the attitude of the technology circle seems to be somewhat Steering. Nvidia CEO Jensen Huang said: "In the future, just leave programming to AI. From now on, everyone will be a software engineer." Not long after, Devin, the world's first AI software engineer, gave The developers brought hundreds of millions of shocks, which really aroused people's heated discussion on the future prospects of the programmer profession. Is it possible that the jobs of programmers are really going to be taken away by AI? In fact, this is not the case. AI is still just a work assistant for programmers. In fact, there is an AI programmer in China who has been working at a major Internet company for some time. It is the new colleague that tens of thousands of Alibaba Cloud engineers have been dealing with frequently recently - "Tongyi Lingma", with the exclusive job number "AI001".
# Before Tongyi Lingma officially joined the company, he first “interned” inside and outside Alibaba Cloud for several months. Everyone's reviews are very good, and the real positive rate is over 80%. In the traditional development model, programmers need to spend a lot of energy writing repetitive code, debugging and optimizing, and writing code comments every day, which greatly squeezes the time for writing core business code. time. But since Tongyi Lingma joined the company, he has played the role of "code assistant" in various development links within Alibaba Cloud, including assisting engineers in the entire process of writing code, reading code, checking bugs, and optimizing code. Moreover, Tongyi Lingma’s work rhythm is even more extreme than 996: on call 24/7... After a few months of internship, Tongyi Lingma has accumulated a lot of "work results". Taking API development and testing work as an example, Tongyi Lingma can shorten the manual test writing time from minutes or even ten minutes to seconds, saving more than 70% of the testing code workload of human programmers. Alibaba Cloud also announced again today that Tongyi Lingma is online on the Tongyi Qianwen APP, which is free and open to everyone. Everyone can write code on their mobile phones anytime, anywhere. , read code, and learn programming skills.
The AI programmer’s skills are maxed out, and he mainly focuses on an all-around assistantIf Tongyi If Lingma has a resume, it must be full of shining points.
Let’s talk about the “background” first. Tongyi Lingma was jointly developed by Alibaba Cloud and Tongyi Lab. Currently, it has more than 2 million downloads and millions of lines every day. Code is adopted by human programmers.
Let’s talk about “professionalism”, Tongyi Lingma has a solid programming foundation, has in-depth study of trillions of code data, and is currently able to use more than 200 programming languages proficiently. , proficient in 16 of them, including Java, Python, Go, JavaScript, TypeScript, C/C, C# and other mainstream languages. When the Yunqi Conference was first released last year, Tongyi Lingma’s many skill points were impressive, such as row-level/function-level real-time continuation, natural language Code generation, unit test generation, code optimization, annotation generation, code explanation and other functions. Give it a piece of project code and let it continue writing code according to the context:
Even unit testing can be done quickly Write it out:
If you are tired of writing code comments, you can let it do it for you:
同時,通靈靈碼也具備研發智能問答、異常報錯排查等能力,減少了原本排查工作的繁雜程度:
此次,正式入職阿里雲的通靈靈碼新增了「程式碼最佳化」功能,升級後的問答模型也更加準確,同時增強了跨文件補全、函數補全等能力,說句“脫胎換骨」不為過。 得益於通靈靈碼的助力,無論是前端、後端還是維運等多個技術職位的員工的工作效率都大大提升,尤其是在編碼單元測試、建構腳本產生和理解、跨語言程式碼轉換等任務上。 通靈靈碼還有一點優勢在於支援企業級程式碼庫專案訓練,能夠更好地理解企業的程式碼和文件數據,並支援國產化適配。未來,阿里雲也將在內部 JetBrains IDEs、Visual Studio Code 等開發工具中全面配置通靈靈碼插件,供所有員工使用。 Talk is cheap,Show me the code說了這麼多,通義靈碼究竟能不能媲美人類程式設計師?還是要看實戰效果。 在開發階段,先看看新增的「程式碼最佳化」能力:還有行/ 函數程式碼的即時續寫: 在「測試」環節,通靈靈碼在產生測試框架、測試案例都非常高效,人類程式設計師需要幾分鐘甚至十幾分鐘的工作,對於它而言只是幾秒鐘的事。 測試架構產生 ## 產生測試案例 此外,通靈靈碼編寫維運腳本的效率也遠高於人類,還可隨時協助尋找維運問題的解決方案:在過程中,開發者有任何問題都可以提問,例如「我想用資料庫來儲存照片,怎麼做」: 當然我們也好奇,為什麼通靈靈碼的程度這麼高?
阿里雲內部的小夥伴給我們透露了幾點:本質上,通靈靈碼能從一個程式設計助手變成一個全能助理,離不開通靈大模型的深厚實力。首先是「基於上下文感知的自適應生成粒度決策」。通義靈碼具備根據語言類型和上下文場景產生正確的程式碼內容(詞級/ 行級/ 區塊級/ 函數級/ 類級等)的能力,既能為開發者產生長程式碼、提高編碼效率,又能適時產生短代碼、減少打擾。 其次是「基於庫內跨檔案自動感知的程式碼產生」。為了解決幻覺問題,通靈靈碼透過先進的端側即時程式碼語義分析技術,即時分析目前正在編寫的程式碼,並基於程式碼語義分析、程式碼引用連結追蹤、動態語言類型推導、相似程式碼分析等先進的技術方法獲取所需的相關代碼引用、相似的代碼片段等語義信息,彌補單純關注當前代碼文件所需的不足。 同時,通靈靈碼能夠「基於檢索增強進行程式碼產生及問答」。具體來說,它透過檢索增強技術即時獲取檢索引擎中的私域知識,讓智慧編碼助理根據企業知識產生程式碼或回答問題,企業可以即時更新及生效私域知識,使其始終基於最新的私域知識生成或回答問題。 讓我們感慨的是,從微軟先前推出的程式設計助理 Copilot 到 AI 程式設計師 Devin,AI 給程式設計師們帶來了一波又一波衝擊。這種革命性的進展,只花了兩年多的時間。 而通靈靈碼在阿里雲正式上崗,標誌著程式設計率先成為大模型技術落地的領域之一。程式語言的邊界從此被打破,Java 程式設計師也能輕鬆完成 Python、C 等不同程式語言的程式碼編寫。問題是,會不會有更多的程式設計師被 AI 端走飯碗? 或許不用太悲觀:正如汽車的出現讓一部分馬車夫沒有了工作,但卻產生了司機這個新的職業。人類程式設計師和 AI 程式設計師的關係,應該是相互促進、共同學習。正如軟體開發方法學的泰山北斗 Kent Beck 所說,有了 AI 工具未來我們 10% 的技能會被 1000 倍的提升。 就比如接下來,迅速融入團隊的通靈靈碼會承擔起更艱鉅的任務,對阿里雲的工程師來說絕對是好消息:「公司未來20% 的程式碼將由通靈靈碼編寫,但程式設計師仍然是研發的核心,他們將有更多時間專注於系統設計以及核心業務開發工作。」對一般人來說,未來也不需要下載專業軟體,使用通義千問就能寫程式碼和學程式碼。針對沒有程式碼基礎的學生,通義千問可以提供學習建議,例如它可以從理論和實操層面幫助使用者制定一套完整學習 Python 的計畫。 The above is the detailed content of The first domestic AI programmer joins Alibaba Cloud: exclusive job number AI001, KPI is that one person can write 20% of the company's code. For more information, please follow other related articles on the PHP Chinese website!