维基百科

人工智能热潮：修订间差异

以互动方式浏览历史
←上一版本
删除的内容 添加的内容
可视化wikitext
无编辑摘要
补救26个来源，并将0个来源标记为失效。) #IABot (v2.0.9.4
 
（未显示3个用户的29个中间版本）
第1行：
{{noteTA
| G1=IT
| G2=PP
}}
{{人工智能}}
'''人工智能热潮'''（{{lang-en|AI boom}}）又称'''人工智能之春'''（{{lang-en|AI spring}}<ref>{{cite web | last=Bommasani | first=Rishi | date=2023-03-17 | url=https://hai.stanford.edu/news/ai-spring-four-takeaways-major-releases-foundation-models | title=AI Spring? Four Takeaways from Major Releases in Foundation Models | work= | publisher=Stanford Institute for Human-Centered Artificial Intelligence | access-date=2023-05-16 | archive-date=2023-05-07 | archive-url=https://web.archive.org/web/20230507165115/https://hai.stanford.edu/news/ai-spring-four-takeaways-major-releases-foundation-models | dead-url=no }}</ref>），指[[人工智能]]领域的快速发展时期。该热潮开始于2016年或2017年的[[OpenAI]]公司，[[生成式人工智能]]为此次热潮的重要组成部分。<ref>{{Cite web |date=2023-03-06 |title=Why am I not terrified of AI? |url=https://scottaaronson.blog/?p=7064 |access-date=2023-03-19 |website=Shtetl-Optimized |language=en-US |archive-date=2023-05-12 |archive-url=https://web.archive.org/web/20230512014057/https://scottaaronson.blog/?p=7064 |dead-url=no }}</ref>OpenAI公司推出的生成式人工智能系统，如2018年推出的各类[[基于转换器的生成式预训练模型|GPT]]以及2021年推出的[[DALL-E]]，皆对此次发展起到了推波助澜的作用。<ref>{{Cite web |last=Newman |first=Daniel |title=Exploring The Ins And Outs Of The Generative AI Boom |url=https://www.forbes.com/sites/danielnewman/2023/03/14/exploring-the-ins-and-outs-of-the-generative-ai-boom/ |access-date=2023-03-14 |website=Forbes |language=en |archive-date=2023-03-28 |archive-url=https://web.archive.org/web/20230328160123/https://www.forbes.com/sites/danielnewman/2023/03/14/exploring-the-ins-and-outs-of-the-generative-ai-boom/ |dead-url=no }}</ref><ref>{{Cite news |date=2023-03-13 |title=The AI boom: lessons from history {{!}} The Economist |newspaper=The Economist |url=https://www.economist.com/finance-and-economics/2023/02/02/the-ai-boom-lessons-from-history |access-date=2023-03-15 |archive-url=https://web.archive.org/web/20230313112611/https://www.economist.com/finance-and-economics/2023/02/02/the-ai-boom-lessons-from-history |archive-date=2023-03-13 }}</ref><ref>{{Cite web |last=Kafka |first=Peter |date=2023-02-01 |title=The AI boom is here, and so are the lawsuits |url=https://www.vox.com/recode/23580554/generative-ai-chatgpt-openai-stable-diffusion-legal-battles-napster-copyright-peter-kafka-column |access-date=2023-03-15 |website=Vox |language=en |archive-date=2023-05-09 |archive-url=https://web.archive.org/web/20230509162659/https://www.vox.com/recode/23580554/generative-ai-chatgpt-openai-stable-diffusion-legal-battles-napster-copyright-peter-kafka-column |dead-url=no }}</ref>
 
由于大型语言模型在2022年得到了提升，基于这一模型的[[聊天机器人]]得以成为现实。与此同时，[[文本到图像生成模型]]所绘图片与手工绘图相比已经到了可以以假乱真的地步<ref>{{Cite web |last=Vincent |first=James |date=2022-05-24 |title=All these images were generated by Google's latest text-to-image AI |url=https://www.theverge.com/2022/5/24/23139297/google-imagen-text-to-image-ai-system-examples-paper |access-date=2023-03-15 |website=The Verge |language=en-US |archive-date=2023-02-15 |archive-url=https://web.archive.org/web/20230215224340/https://www.theverge.com/2022/5/24/23139297/google-imagen-text-to-image-ai-system-examples-paper |dead-url=no }}</ref>，而[[语音合成]]技术亦可成功的模仿人类讲话。<ref>{{Cite web |title=AI-Generated Voice Firm Clamps Down After 4chan Makes Celebrity Voices for Abuse |url=https://www.vice.com/en/article/dy7mww/ai-voice-firm-4chan-celebrity-voices-emma-watson-joe-rogan-elevenlabs |access-date=2023-03-15 |website=www.vice.com |language=en |archive-date=2023-05-07 |archive-url=https://web.archive.org/web/20230507055208/https://www.vice.com/en/article/dy7mww/ai-voice-firm-4chan-celebrity-voices-emma-watson-joe-rogan-elevenlabs |dead-url=no }}</ref>
 
自2022年底至2023年间，随着科技巨擘在市场上站稳脚跟，数十个新AI网站和AI聊天机器人上线，导致AI工具的普及率空前增加。<ref>{{cite news |last1=Firth-Butterfield |first1=Kay |title=2022 was a big year for AI development. In 2023, we must decide how best to use it |url=https://asianews.network/2022-was-a-big-year-for-ai-development-in-2023-we-must-decide-how-best-to-use-it/ |access-date=2023-05-16 |work=Asia News Network |date=2023-01-18 |archive-date=2023-03-19 |archive-url=https://web.archive.org/web/20230319152212/https://asianews.network/2022-was-a-big-year-for-ai-development-in-2023-we-must-decide-how-best-to-use-it/ |dead-url=no }}</ref>对于这一AI热潮，民众对此的反应可谓褒贬不一。有的人认为AI有助于开发人类潜力，为人类谋福祉。但有的人则认为AI会导致大量失业，并指出了目前的AI技术所存在诸多缺陷。<ref>{{cite web |title=No matter how sophisticated, artificial intelligence systems still need human oversight |url=https://www.zdnet.com/article/no-matter-how-sophisticated-artificial-intelligence-systems-still-need-the-human-touch/ |website=ZDNET |access-date=2023-05-16 |language=en |archive-date=2023-05-10 |archive-url=https://web.archive.org/web/20230510092604/https://www.zdnet.com/article/no-matter-how-sophisticated-artificial-intelligence-systems-still-need-the-human-touch/ |dead-url=no }}</ref><ref>{{cite web |last1=Sukhadeve |first1=Ashish |title=Council Post: Artificial Intelligence For Good: How AI Is Helping Humanity |url=https://www.forbes.com/sites/forbesbusinesscouncil/2021/02/09/artificial-intelligence-for-good-how-ai-is-helping-humanity/?sh=235475b03366 |website=Forbes |access-date=2023-05-16 |language=en |archive-date=2023-05-09 |archive-url=https://web.archive.org/web/20230509052233/https://www.forbes.com/sites/forbesbusinesscouncil/2021/02/09/artificial-intelligence-for-good-how-ai-is-helping-humanity/?sh=235475b03366 |dead-url=no }}</ref><ref>{{cite web |title=Could AI advancements be a threat to your job security? {{!}} Learning People |url=https://www.learningpeople.com/uk/blog/career-advice/7-reasons-why-now-is-the-perfect-time-for-a-career-in-digital-marketing-0-0/ |website=www.learningpeople.com |access-date=2023-05-16 |archive-date=2023-05-09 |archive-url=https://web.archive.org/web/20230509061107/https://www.learningpeople.com/uk/blog/career-advice/7-reasons-why-now-is-the-perfect-time-for-a-career-in-digital-marketing-0-0/ |dead-url=no }}</ref><ref>{{cite web |last1=Zinkula |first1=Aaron Mok, Jacob |title=ChatGPT may be coming for our jobs. Here are the 10 roles that AI is most likely to replace. |url=https://www.businessinsider.com/chatgpt-jobs-at-risk-replacement-artificial-intelligence-ai-labor-trends-2023-02 |website=Business Insider |access-date=2023-05-16 |archive-date=2023-05-09 |archive-url=https://web.archive.org/web/20230509073337/https://www.businessinsider.com/chatgpt-jobs-at-risk-replacement-artificial-intelligence-ai-labor-trends-2023-02 |dead-url=no }}</ref>
== 语言模型 ==
[[GPT-3]]是OpenAI于2020年发布的[[大型语言模型]]，该模型可以生成高质量的类人文本，其品质几可乱真。<ref>{{Cite web |last=Sagar |first=Ram |date=2020-06-03 |title=OpenAI Releases GPT-3, The Largest Model So Far |url=https://analyticsindiamag.com/open-ai-gpt-3-language-model/ |access-date=2023-03-15 |website=Analytics India Magazine |language=en-US |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>随后OpenAI又发布了GPT-3.5版本，该版本被用于[[ChatGPT]]，因其能对多个领域的问题给出清晰回答而广受关注。<ref>{{Cite news |last=Lock |first=Samantha |date=2022-12-05 |title=What is AI chatbot phenomenon ChatGPT and could it replace humans? |language=en-GB |work=The Guardian |url=https://www.theguardian.com/technology/2022/dec/05/what-is-ai-chatbot-phenomenon-chatgpt-and-could-it-replace-humans |access-date=2023-03-15 |issn=0261-3077 |archive-date=2023-01-16 |archive-url=https://web.archive.org/web/20230116100346/https://www.theguardian.com/technology/2022/dec/05/what-is-ai-chatbot-phenomenon-chatgpt-and-could-it-replace-humans |dead-url=no }}</ref>随后OpenAI又于2023年4月发布了[[GPT-4]]模型，目前该模型已被用于[[Microsoft Bing]]搜索引擎。<ref>{{Cite web |last=Lardinois |first=Frederic |date=2023-03-14 |title=Microsoft's new Bing was using GPT-4 all along |url=https://techcrunch.com/2023/03/14/microsofts-new-bing-was-using-gpt-4-all-along/ |access-date=2023-03-15 |website=TechCrunch |language=en-US |archive-date=2023-03-15 |archive-url=https://web.archive.org/web/20230315013650/https://techcrunch.com/2023/03/14/microsofts-new-bing-was-using-gpt-4-all-along/ |dead-url=no }}</ref><ref>{{Cite news |date=2023-03-14 |title=OpenAI announces ChatGPT successor GPT-4 |language=en-GB |work=BBC News |url=https://www.bbc.com/news/technology-64959346 |access-date=2023-03-15 |archive-date=2023-05-15 |archive-url=https://web.archive.org/web/20230515221328/https://www.bbc.com/news/technology-64959346 |dead-url=no }}</ref>除了OpenAI以外，还有其他多个公司皆发布了自己的语言模型，如[[谷歌]]发布的[[LaMDA]]模型以及[[Meta]]发布的[[LLaMA]]模型。
== 文本到图像生成模型 ==
[[OpenAI]]发布的[[DALL-E]]为首批受广泛关注的[[文本到图像生成模型]]之一，该模型发布于2021年1月。<ref>{{Cite web |last=Coldewey |first=Devin |date=2021-01-05 |title=OpenAI's DALL-E creates plausible images of literally anything you ask it to |url=https://techcrunch.com/2021/01/05/openais-dall-e-creates-plausible-images-of-literally-anything-you-ask-it-to/ |access-date=2023-03-15 |website=TechCrunch |language=en-US |archive-date=2021-01-06 |archive-url=https://web.archive.org/web/20210106075542/https://techcrunch.com/2021/01/05/openais-dall-e-creates-plausible-images-of-literally-anything-you-ask-it-to/ |dead-url=no }}</ref>随后能够生成更为逼真图像的DALL-E 2于2022年4月发布，<ref>{{Cite web |last=Coldewey |first=Devin |date=2022-04-06 |title=New OpenAI tool draws anything, bigger and better than ever |url=https://techcrunch.com/2022/04/06/openais-new-dall-e-model-draws-anything-but-bigger-better-and-faster-than-before/ |access-date=2023-03-15 |website=TechCrunch |language=en-US |archive-date=2023-05-06 |archive-url=https://web.archive.org/web/20230506073242/https://techcrunch.com/2022/04/06/openais-new-dall-e-model-draws-anything-but-bigger-better-and-faster-than-before/ |dead-url=no }}</ref>而其开源替代版本[[Stable Diffusion]]则发布于2022年8月。<ref>{{Cite web |title=Stable Diffusion Public Release |url=https://stability.ai/blog/stable-diffusion-public-release |access-date=2023-03-15 |website=Stability AI |language=en-GB |archive-date=2022-08-30 |archive-url=https://web.archive.org/web/20220830210535/https://stability.ai/blog/stable-diffusion-public-release |dead-url=no }}</ref>
 
继文本到图像生成模型之后，又涌现了多个受语言模型驱动的文本到视频生成模型，如DAMO、<ref>{{Cite web |title=ModelScope 魔搭社区 |url=https://modelscope.cn/models/damo/text-to-video-synthesis/summary |access-date=2023-03-20 |website=modelscope.cn |archive-date=2023-05-09 |archive-url=https://web.archive.org/web/20230509191246/https://modelscope.cn/models/damo/text-to-video-synthesis/summary |dead-url=no }}</ref> Make-A-Video、<ref>{{Cite web |last=kumar |first=Ashish |date=2022-10-03 |title=Meta AI Introduces 'Make-A-Video': An Artificial Intelligence System That Generates Videos From Text |url=https://www.marktechpost.com/2022/10/03/meta-ai-introduces-make-a-video-an-artificial-intelligence-system-that-generates-videos-from-text/ |access-date=2023-03-15 |website=MarkTechPost |language=en-US |archive-date=2022-12-01 |archive-url=https://web.archive.org/web/20221201101936/https://www.marktechpost.com/2022/10/03/meta-ai-introduces-make-a-video-an-artificial-intelligence-system-that-generates-videos-from-text/ |dead-url=no }}</ref>Imagen Video<ref>{{Cite web |last=Edwards |first=Benj |date=2022-10-05 |title=Google's newest AI generator creates HD video from text prompts |url=https://arstechnica.com/information-technology/2022/10/googles-newest-ai-generator-creates-hd-video-from-text-prompts/ |access-date=2022-10-25 |website=Ars Technica |language=en-us |archive-date=2023-02-07 |archive-url=https://web.archive.org/web/20230207024732/https://arstechnica.com/information-technology/2022/10/googles-newest-ai-generator-creates-hd-video-from-text-prompts/ |dead-url=no }}</ref>以及Phenaki<ref>{{Cite web |title=Phenaki |url=https://phenaki.video/?mc_cid=9fee7eeb9d&mc_eid=6e7303fddd#interactive |access-date=2022-10-03 |website=phenaki.video |archive-date=2022-10-07 |archive-url=https://web.archive.org/web/20221007071332/https://phenaki.video/?mc_cid=9fee7eeb9d&mc_eid=6e7303fddd#interactive |dead-url=no }}</ref>，这些模型可通过文本或文本/图像提示自动生成视频。<ref>{{cite news |last1=Edwards |first1=Benj |date=2022-09-09 |title=Runway teases AI-powered text-to-video editing using written prompts |publisher=Ars Technica |url=https://arstechnica.com/information-technology/2022/09/runway-teases-ai-powered-text-to-video-editing-using-written-prompts/ |access-date=2022-09-12 |archive-date=2023-01-27 |archive-url=https://web.archive.org/web/20230127024508/https://arstechnica.com/information-technology/2022/09/runway-teases-ai-powered-text-to-video-editing-using-written-prompts/ |dead-url=no }}</ref>
== 语音合成 ==
[[15.ai]]是首批开放使用的语言合成软件，该产品发布于2020年3月，能够让人们通过输入音频媒体的方式产生对其声音的模仿。<ref name="kotaku">{{cite web |last=Zwiezen |first=Zack |date=2021-01-18 |title=Website Lets You Make GLaDOS Say Whatever You Want |url=https://kotaku.com/this-website-lets-you-make-glados-say-whatever-you-want-1846062835 |url-status=live |archive-url=https://web.archive.org/web/20210117164748/https://kotaku.com/this-website-lets-you-make-glados-say-whatever-you-want-1846062835 |archive-date=2021-01-17 |access-date=2021-01-18 |website=[[Kotaku]] |publisher=[[Kotaku]] |quote=}}</ref><ref name="gameinformer">{{cite magazine |last=Ruppert |first=Liana |date=2021-01-18 |title=Make Portal's GLaDOS And Other Beloved Characters Say The Weirdest Things With This App |url=https://www.gameinformer.com/gamer-culture/2021/01/18/make-portals-glados-and-other-beloved-characters-say-the-weirdest-things |url-status=live |magazine=[[Game Informer]] |publisher=[[Game Informer]] |archive-url=https://web.archive.org/web/20210118175543/https://www.gameinformer.com/gamer-culture/2021/01/18/make-portals-glados-and-other-beloved-characters-say-the-weirdest-things |archive-date=2021-01-18 |access-date=2021-01-18 |quote=}}</ref>随后ElevenLabs推出了一个允许公众上传其声音的网站，用于语音合成。不过由于该软件能够让用户模仿知名人士的声音并利用此发布假声名，ElevenLabs受到了各界谴责。<ref>{{Cite news |author1=Jorge Jimenez |date=2023-01-31 |title=AI company promises changes after 'voice cloning' tool used to make celebrities say awful things |language=en |work=PC Gamer |url=https://www.pcgamer.com/ai-company-promises-changes-after-voice-cloning-tool-used-to-make-celebrities-say-awful-things/ |access-date=2023-02-03 |archive-date=2023-04-04 |archive-url=https://web.archive.org/web/20230404164634/https://www.pcgamer.com/ai-company-promises-changes-after-voice-cloning-tool-used-to-make-celebrities-say-awful-things/ |dead-url=no }}</ref>除此之外，这一技术引发了民众对其应用于[[深伪技术|深度伪造]]的担忧。<ref>{{cite web |title=Seeing is believing? Global scramble to tackle deepfakes |url=https://news.yahoo.com/seeing-believing-global-scramble-tackle-013429757.html |website=Yahoo News |access-date=2023-05-16 |archive-date=2023-02-03 |archive-url=https://web.archive.org/web/20230203073245/https://news.yahoo.com/seeing-believing-global-scramble-tackle-013429757.html |dead-url=no }}</ref>在有人利用语音合成技术对[[德雷克 (歌手)|德雷克]]与[[威肯]]的声音进行音乐创作后，不少民众对该技术的合法性及道德提出了质疑。<ref>{{Cite news |last=Coscarelli |first=Joe |date=2023-04-19 |title=An A.I. Hit of Fake ‘Drake’ and ‘The Weeknd’ Rattles the Music World |language=en-US |work=The New York Times |url=https://www.nytimes.com/2023/04/19/arts/music/ai-drake-the-weeknd-fake.html |access-date=2023-05-16 |issn=0362-4331 |archive-date=2023-05-15 |archive-url=https://web.archive.org/web/20230515182004/https://www.nytimes.com/2023/04/19/arts/music/ai-drake-the-weeknd-fake.html |dead-url=no }}</ref>
 
== 参考文献 ==
{{reflist|30em}}
 
{{OpenAI navbox}}
 
[[Category:人工智能]]
检索自“https://zh.wikipedia.org/wiki/人工智能热潮”