imported>Yumeto：使用DisamAssist清理消歧義連結：Go（改連結至Go語言）。

2026-03-02T15:27:25Z

新页面

{{Request translation|en|time=2022-08-04}}
{{Distinguish|Microsoft Copilot|Microsoft 365 Copilot}}
{{NoteTA
|G1=IT
}}
{{infobox software
| logo = GitHub Copilot logo.svg
| developer = [[GitHub]]、[[OpenAI]]
| operating system = [[Microsoft Windows]]、[[Linux]]、[[macOS]]、Web
| website = {{URL|https://github.com/features/copilot/}}
}}
'''GitHub Copilot'''是[[GitHub]]和[[OpenAI]]合作开发的一个[[人工智能]]工具，用户在使用[[Visual Studio Code]]、[[Microsoft Visual Studio]]、[[Vim]]、{{link-en|Cursor|Cursor (code editor)}}或[[JetBrains]][[集成开发环境]]時可以通過GitHub Copilot[[自动完成|自动补全]]-{zh-hans:代码; zh-hant:程式碼;}-<ref name=":0">{{Cite web|title=GitHub and OpenAI launch a new AI tool that generates its own code|url=https://www.theverge.com/2021/6/29/22555777/github-openai-ai-tool-autocomplete-code|access-date=2021-07-06|author=Gershgorn|date=2021-06-29|first=Dave|work=[[The Verge]]|archive-date=2021-07-07|archive-url=https://web.archive.org/web/20210707105519/https://www.theverge.com/2021/6/29/22555777/github-openai-ai-tool-autocomplete-code}}</ref>。GitHub于2021年6月29日對外公開该软件<ref name="Verge legal">{{Cite news|url=https://www.theverge.com/2021/7/7/22561180/github-copilot-legal-copyright-fair-use-public-code|title=GitHub’s automatic coding tool rests on untested legal ground|date=2021-07-07|accessdate=2021-07-11|work=The Verge|archive-date=2021-07-11|archive-url=https://web.archive.org/web/20210711071654/https://www.theverge.com/2021/7/7/22561180/github-copilot-legal-copyright-fair-use-public-code}}</ref>，GitHub Copilot於[[軟件版本週期|技术预览阶段]]主要面向[[Python]]、[[JavaScript]]、[[TypeScript]]、[[Ruby]]和[[Go語言|Go]]等编程语言<ref name="Verge legal" /><ref name=":2">{{Cite web |title=GitHub Copilot · Your AI pair programmer |url=https://copilot.github.com/ |access-date=7 April 2022 |website=GitHub Copilot |archive-date=2022-06-19 |archive-url=https://web.archive.org/web/20220619101317/https://copilot.github.com/ |dead-url=no }}</ref>。

== 历史 ==
2021年6月29日，GitHub 宣布 GitHub Copilot 在 Visual Studio Code 开发环境中进行技术预览。<ref name=":0" /><ref>{{Cite web |date=2021-06-29 |title=Introducing GitHub Copilot: your AI pair programmer |url=https://github.blog/2021-06-29-introducing-github-copilot-ai-pair-programmer/ |website=The GitHub Blog |language=en-US |access-date=2022-04-07 |archive-date=2022-06-21 |archive-url=https://web.archive.org/web/20220621205502/https://github.blog/2021-06-29-introducing-github-copilot-ai-pair-programmer/ |dead-url=no }}</ref>

2021年10月26日，GitHub Copilot 作为[[插件]]在 JetBrains 市场上发布。<ref>{{Cite web |title=GitHub Copilot - IntelliJ IDEs Plugin {{!}} Marketplace |url=https://plugins.jetbrains.com/plugin/17718-github-copilot/versions/stable |website=JetBrains Marketplace |access-date=2022-04-07 |archive-date=2022-06-27 |archive-url=https://web.archive.org/web/20220627073722/https://plugins.jetbrains.com/plugin/17718-github-copilot/versions/stable |dead-url=no }}</ref>

2021年10月27日，GitHub 发布了 GitHub Copilot Neovim 插件作为公共存储库。<ref>{{Citation|title=Copilot.vim|date=7 April 2022|url=https://github.com/github/copilot.vim|publisher=GitHub|access-date=7 April 2022|archive-date=2022-06-13|archive-url=https://web.archive.org/web/20220613124243/https://github.com/github/copilot.vim|dead-url=no}}</ref>

2022年3月9日，Github官方宣布Copilot适用于Visual Studio 2022 IDE。<ref>{{Cite web |date=29 March 2022 |title=GitHub Copilot now available for Visual Studio 2022 |url=https://github.blog/2022-03-29-github-copilot-now-available-for-visual-studio-2022/ |website=The GitHub Blog |language=en-US |access-date=7 April 2022 |archive-date=2022-06-14 |archive-url=https://web.archive.org/web/20220614172855/https://github.blog/2022-03-29-github-copilot-now-available-for-visual-studio-2022/ |dead-url=no }}</ref>

2022年6月21日，GitHub 正式宣布 Copilot 退出“技术预览版”，并作为基于订阅的服务提供给个人开发者。<ref>{{Cite web |date=21 June 2022 |title=GitHub Copilot is generally available to all developers |url=https://github.blog/2022-06-21-github-copilot-is-generally-available-to-all-developers/ |website=The GitHub Blog |language=en-US |access-date=21 June 2022 |archive-date=2022-06-23 |archive-url=https://web.archive.org/web/20220623002316/https://github.blog/2022-06-21-github-copilot-is-generally-available-to-all-developers/ |dead-url=no }}</ref>此外，GitHub 提到 Copilot 将继续“对经过验证的学生和流行开源项目的维护者免费”，并且在2022年晚些时候提供给公司。

== 特性 ==
GitHub Copilot 由 [[OpenAI Codex]] 提供支持，OpenAI Codex 是由人工智能研究实验室 OpenAI 创建的人工智能模型。<ref>{{Cite web |last=Krill |first=Paul |date=12 August 2021 |title=OpenAI offers API for GitHub Copilot AI model |url=https://www.infoworld.com/article/3629469/openai-offers-api-for-github-copilot-ai-model.html |website=InfoWorld |language=en |access-date=7 April 2022 |archive-date=2022-04-17 |archive-url=https://web.archive.org/web/20220417150725/https://www.infoworld.com/article/3629469/openai-offers-api-for-github-copilot-ai-model.html |dead-url=no }}</ref>OpenAI Codex 是 [[GPT-3]] 的修改后生产版本，GPT-3 是一种使用[[深度学习]]生成类人类文本的语言模型。<ref>{{Cite web |date=3 June 2020 |title=OpenAI Releases GPT-3, The Largest Model So Far |url=https://analyticsindiamag.com/open-ai-gpt-3-language-model/ |website=Analytics India Magazine |language=en-US |access-date=7 April 2022 |archive-date=2020-08-04 |archive-url=https://web.archive.org/web/20200804173452/https://analyticsindiamag.com/open-ai-gpt-3-language-model/ |dead-url=no }}</ref>例如，当给出一个[[自然语言]]的程序问题时，Codex能够产生解法代码。<ref name=":1">{{Cite journal |last1=Finnie-Ansley |first1=James |last2=Denny |first2=Paul |last3=Becker |first3=Brett A. |last4=Luxton-Reilly |first4=Andrew |last5=Prather |first5=James |date=14 February 2022 |title=The Robots Are Coming: Exploring the Implications of OpenAI Codex on Introductory Programming |url=https://doi.org/10.1145/3511861.3511863 |journal=Australasian Computing Education Conference |series=ACE '22 |location=New York, NY, USA |publisher=Association for Computing Machinery |pages=10–19 |doi=10.1145/3511861.3511863 |isbn=978-1-4503-9643-1 |s2cid=246681316}}</ref>它也可以用[[英语]]描述输入代码和在不同程序语言之间翻译代码。<ref name=":1" />Codex的GPT-3仅授权给 GitHub 的[[母公司]][[微软]]。<ref>{{Cite web |title=OpenAI is giving Microsoft exclusive access to its GPT-3 language model |url=https://www.technologyreview.com/2020/09/23/1008729/openai-is-giving-microsoft-exclusive-access-to-its-gpt-3-language-model/ |website=MIT Technology Review |language=en |access-date=7 April 2022 |archive-date=2021-02-05 |archive-url=https://web.archive.org/web/20210205121656/https://www.technologyreview.com/2020/09/23/1008729/openai-is-giving-microsoft-exclusive-access-to-its-gpt-3-language-model/ |dead-url=no }}</ref>

Copilot的OpenAI Codex接受了一系列英语语言、公共 GitHub 存储库和其他公开可用源代码的训练。<ref name=":2" />这包括来自 5400 万个公共 GitHub 存储库的 159 [[吉字节|GB]] Python 代码的过滤数据集。<ref>{{Cite web |title=OpenAI Announces 12 Billion Parameter Code-Generation AI Codex |url=https://www.infoq.com/news/2021/08/openai-codex/ |website=InfoQ |language=en |access-date=7 April 2022 |archive-date=2022-07-09 |archive-url=https://web.archive.org/web/20220709221205/https://www.infoq.com/news/2021/08/openai-codex/ |dead-url=no }}</ref>

根据它的网站，GitHub Copilot包括协助程序员的功能，如从[[註解 (程式設計)|代码注释]]到可运行代码的转换，和自动补全代码块、重复的代码、以及整个[[方法 (计算机科学)|方法]]或[[子程序|函数]]。<ref name=":2" /><ref>{{Cite journal |last1=Sobania |first1=Dominik |last2=Schweim |first2=Dirk |last3=Rothlauf |first3=Franz |date=2022 |title=A Comprehensive Survey on Program Synthesis with Evolutionary Algorithms |url=https://ieeexplore.ieee.org/abstract/document/9743417?casa_token=2vffB_1z0UsAAAAA:7YGw6NRDYQE9UBDPUdv7Ij7EwySLOMENhzzRmIYo5erRHv0qtESQn-lCkMAh35cWn-DVrcP69Q |journal=IEEE Transactions on Evolutionary Computation |pages=1 |doi=10.1109/TEVC.2022.3162324 |issn=1941-0026 |s2cid=247721793 |access-date=2022-08-04 |archive-date=2022-04-01 |archive-url=https://web.archive.org/web/20220401025328/https://ieeexplore.ieee.org/abstract/document/9743417?casa_token=2vffB_1z0UsAAAAA:7YGw6NRDYQE9UBDPUdv7Ij7EwySLOMENhzzRmIYo5erRHv0qtESQn-lCkMAh35cWn-DVrcP69Q |dead-url=no }}</ref>GitHub 报告称，Copilot 的自动完成功能大约有一半时间是准确的；例如，當用戶提供Python 函数头代码後，Copilot 在第一次尝试时有 43% 的时间正确地自动完成了函数体代码的其余部分，而在十次尝试后有 57% 的时间正确地自动完成了函数体代码的其余部分。<ref name=":2" />

GitHub 表示，Copilot 的功能能夠幫助程序员减省阅读[[软件文档]]的时间，快速瀏覽不熟悉的编码[[软件框架|框架]]和语言。<ref name=":2" />

== 各界反应 ==

=== 许可证争议 ===
GitHub承认，该工具在输出部分代码时会逐句复制，这会导致对输出的代码是否属于合理使用的争议，并可能侵犯原始所有者的版权。<ref>{{Cite web|last=Gershgorn|first=Dave|date=2021-07-07|title=Can AI learn from any public code online?|url=https://www.theverge.com/2021/7/7/22561180/github-copilot-legal-copyright-fair-use-public-code|access-date=2023-03-23|website=The Verge|language=en-US|archive-date=2021-07-11|archive-url=https://web.archive.org/web/20210711071654/https://www.theverge.com/2021/7/7/22561180/github-copilot-legal-copyright-fair-use-public-code|dead-url=no}}</ref>2022年6月，[[软件自由保护组织]]宣布将停止在自己的项目中使用GitHub服务，<ref>{{Cite web|title=Give Up GitHub: The Time Has Come!|url=https://sfconservancy.org/blog/2022/jun/30/give-up-github-launch/|access-date=2023-03-23|website=Software Freedom Conservancy|language=en|archive-date=2023-05-24|archive-url=https://web.archive.org/web/20230524125119/https://sfconservancy.org/blog/2022/jun/30/give-up-github-launch/|dead-url=no}}</ref>并控诉GitHub在训练代码时忽视软件许可证。<ref>{{Cite web|title=If Software is My Copilot, Who Programmed My Software?|url=https://sfconservancy.org/blog/2022/feb/03/github-copilot-copyleft-gpl/|access-date=2023-03-23|website=Software Freedom Conservancy|language=en|archive-date=2023-05-27|archive-url=https://web.archive.org/web/20230527230845/https://sfconservancy.org/blog/2022/feb/03/github-copilot-copyleft-gpl/|dead-url=no}}</ref>

=== FSF白皮书 ===
2021年7月28日，[[自由软件基金会]]（FSF）发表了一篇呼吁资助来探讨GitHub Copilot相关哲学与法律问题的白皮书。<ref>{{Cite web|title=FSF-funded call for white papers on philosophical and legal questions around Copilot: Submit before Monday, August 23, 2021 — Free Software Foundation — Working together for free software|url=https://www.fsf.org/blogs/licensing/fsf-funded-call-for-white-papers-on-philosophical-and-legal-questions-around-copilot|access-date=2023-03-23|website=www.fsf.org|archive-date=2021-08-11|archive-url=https://web.archive.org/web/20210811003717/https://www.fsf.org/blogs/licensing/fsf-funded-call-for-white-papers-on-philosophical-and-legal-questions-around-copilot|dead-url=no}}</ref>

=== 隐私问题 ===
GitHub Copilot是云计算服务，需要持续和GitHub Copilot服务器通讯以正常使用。<ref>{{Cite web|title=GitHub Copilot · Your AI pair programmer|url=https://github.com/features/copilot/|access-date=2023-03-23|website=GitHub|language=en|archive-date=2023-06-11|archive-url=https://web.archive.org/web/20230611164114/https://github.com/features/copilot|dead-url=no}}</ref>这种不透明的架构引发了对数据挖掘和按键遥测的担忧。<ref>{{Cite web|title=CoPilot: Privacy & DataMining · community/community · Discussion #7263|url=https://github.com/orgs/community/discussions/7263|access-date=2023-03-23|website=GitHub|language=en|archive-date=2023-02-21|archive-url=https://web.archive.org/web/20230221211421/https://github.com/orgs/community/discussions/7263|dead-url=no}}</ref><ref>{{Cite web|title=Who Does That Server Really Serve? - GNU Project - Free Software Foundation|url=https://www.gnu.org/philosophy/who-does-that-server-really-serve.en.html|access-date=2023-03-23|website=www.gnu.org|archive-date=2023-06-09|archive-url=https://web.archive.org/web/20230609163535/https://www.gnu.org/philosophy/who-does-that-server-really-serve.en.html|dead-url=no}}</ref>

=== 教育问题 ===
[[计算机协会]]于2022年8月发布的一篇论文评估了GitHub Copilot使用的技术Codex可能会对新手程序员的教育产生的影响。<ref>{{Cite journal |last=Finnie-Ansley |first=James |last2=Denny |first2=Paul |last3=Becker |first3=Brett A. |last4=Luxton-Reilly |first4=Andrew |last5=Prather |first5=James |date=2022-02-14 |title=The Robots Are Coming: Exploring the Implications of OpenAI Codex on Introductory Programming |url=https://dl.acm.org/doi/10.1145/3511861.3511863 |journal=Proceedings of the 24th Australasian Computing Education Conference |series=ACE '22 |location=New York, NY, USA |publisher=Association for Computing Machinery |doi=10.1145/3511861.3511863 |isbn=978-1-4503-9643-1 |access-date=2025-01-14 |archive-date=2025-01-14 |archive-url=https://web.archive.org/web/20250114053338/https://dl.acm.org/doi/10.1145/3511861.3511863 |dead-url=no }}</ref>

== 相關條目 ==
* [[ChatGPT]]
* [[智能代码补全]]
* {{link-en|Devin AI|Devin AI}}
* [[生成式人工智慧]]
* [[Microsoft Copilot]]
* [[Vibe coding]]
==参考文献==
{{reflist|30em}}

{{OpenAI}}
{{Differentiable computing}}
{{生成式人工智慧}}

[[Category:人工智能应用]]
[[Category:GitHub]]
[[Category:OpenAI]]

GitHub Copilot - 版本历史

imported>Yumeto：​使用DisamAssist清理消歧義連結：Go（改連結至Go語言）。

imported>Yumeto：使用DisamAssist清理消歧義連結：Go（改連結至Go語言）。