How Important is Confidentiality in Translation Services?

翻译服务中的保密性有多重要?

2024-08-30 05:58 slator

本文共832个字,阅读需9分钟

阅读模式 切换至中文

In 2023, the Hollywood actor strike that paralyzed a number of localization activities in film, TV, and gaming productions, among others, allowed for dubbing work to continue. The July 2024 SAG–AFTRA union’s strike would require members to stop working altogether in video games, specifically, as covered by the union’s “Interactive Media Agreement.” “Localization for foreign video games covered under the Interactive Localization Agreement (ILA) is struck work,” says a statement published by the union. The list of services becoming off limits to members includes acting; singing; “voice acting, including performing sound-alike voice services;” and “authorizing the use of your voice or likeness (which includes integration or reuse of work already performed)” in video games. The latest strike highlights a key point of contention: the use of AI in voice acting. SAG-AFTRA aims to protect its members from being replaced by AI technology, just as game studios are ever more tempted by the potential of AI dubbing to lower costs and shorten production timelines. We asked readers if they think the video game actor strike will be a net positive for voice actors, and over a third (36.8%) believe that will be the case. A group of about the same size (34.2%) thinks it is probable, while less than a quarter (23.7%) find it unlikely and the rest (5.3%) think it will not be so. “Iyuno is aware of a recent security issue, involving unauthorized access to confidential content. Protecting our clients’ confidentiality and ensuring the security of their content is our highest priority,” read the statement made by the media localization company about a security breach resulting in the leak of unreleased content. According to a “What’s on Netflix” news article, the Netflix post-production partner was the target of the hacking of the localized versions of “at least nine Netflix original series and movies.” Netflix apparently got to work immediately to remove the leaked content and secure its systems, but not before clips, footage, and complete episodes of several series had made their way to social networks and torrent sites. Some in the anime community voiced their discontent about the leak on social media, begging the hackers and takers to stop sharing. We wanted to know how important readers think confidentiality is in their area of language services and technology as the Iyuno content leak brings to the forefront the vulnerabilities of content security for language service providers (LSPs). The vast majority of respondents (73.3%) think it is absolutely critical. Less than a quarter (20.0%) of readers consider confidentiality important, and a small group (6.7%) does not see it as important. Language AI, especially machine translation (MT) tools, is reaching higher levels of democratization. People from all backgrounds can now perform multilingual speech and text conversion and generation using Internet sites and consumer grade applications. Language AI meant for consumers is bringing technologies like automated captions, machine dubbing, speech translation, and AI text generation to desktops and mobile devices, including multilingual communications and web plugins. Not too concerned about end quality, these consumers are mostly trying to communicate in other languages as best as the AI allows, usually making it clear that it was AI that translated the text or voice. But before consumers could easily use language AI, individual content creators were already experimenting with it in commercial ways, allowing them to reach broader audiences on video platforms and websites. Things like automated transcription services, subtitles, and captions are increasingly affordable and accessible. Many are also free, up to certain volumes. Language AI is expanding localization capabilities at unprecedented levels for everyone, so we asked Slator readers to share whether their friends and acquaintances (outside the language industry) use language AI technologies. Respondents believe that a little over a third (35.4%) use AI sometimes, and another third (31.3%) use it often. They also find that about a quarter have probably tried it out (27.1%), and the rest (6.2%) have never tried it out. An analysis of articles on the AI vs human translators topic conducted by Slator in July 2024 highlighted some recurrent themes, with about a quarter of the sampled publications mentioning human translators as “indispensable.” Citing concerns about losing the human touch in creative, culturally accurate translations as AI translation becomes pervasive, a significant portion of media coverage praises the abilities of human translators, contrasting them with obvious limitations of AI such as hallucinations and use of incorrect terms. A larger segment of coverage (40%) portrays the decline of translators' livelihoods as something inevitable, citing job losses and layoffs due to AI. The narrative tends to use some familiar qualifiers, such as declining demand for human translation, AI as a threat to translator jobs, and even speculations about the end of foreign language education. We asked readers how accurate they think translation-related AI coverage of major media outlets is, and over a third (36.8%) answered that it is not very accurate. Two groups are almost evenly split between those who believe it is somewhat accurate (23.7%) or very accurate (21.1%), and the rest find it to be consistently wrong (18.4%).
二零二三年,好莱坞演员罢工导致电影、电视及游戏制作等多项本地化活动陷入瘫痪,配音工作得以继续。2024年7月SAG-AFTRA工会的罢工将要求成员完全停止在视频游戏中的工作,具体来说,正如工会的“互动媒体协议”所涵盖的那样。 该联盟发表的一份声明说:“互动本地化协议(ILA)所涵盖的外国视频游戏的本地化是一项艰巨的工作。” 禁止会员使用的服务包括表演、唱歌、“配音,包括执行声音相似的语音服务”以及“授权在视频游戏中使用您的声音或肖像(包括集成或重复使用已执行的工作)”。 最新的罢工突出了一个关键的争论点:人工智能在配音中的使用。SAG-AFTRA旨在保护其成员免受人工智能技术的取代,就像游戏工作室越来越受到人工智能配音降低成本和缩短制作时间表的潜力的诱惑一样。 我们询问读者,他们是否认为视频游戏演员罢工对配音演员来说是一个净积极因素,超过三分之一(36.8%)的人认为会是这样。大约相同规模的群体(34.2%)认为这是可能的,而不到四分之一(23.7%)认为这是不可能的,其余(5.3%)认为不会如此。 “Iyuno意识到最近的一个安全问题,涉及未经授权访问机密内容。保护我们客户的机密性并确保其内容的安全性是我们的首要任务,”该媒体本地化公司就导致未发布内容泄露的安全漏洞发表声明。 根据“Netflix上有什么”的新闻文章,Netflix的后期制作合作伙伴是“至少九部Netflix原创剧集和电影”的本地化版本的黑客攻击目标。Netflix显然立即着手删除泄露的内容并保护其系统,但在此之前,几部剧集的片段、镜头和完整剧集已经进入了社交网络和洪流网站。 一些动漫社区的人在社交媒体上表达了他们对泄密的不满,恳求黑客和接受者停止分享。 我们想知道读者认为保密性在他们的语言服务和技术领域有多重要,因为Iyuno内容泄漏将语言服务提供商(LSP)的内容安全漏洞推到了最前沿。绝大多数受访者(73.3%)认为这是绝对关键的。 不到四分之一(20.0%)的读者认为保密性很重要,还有一小部分(6.7%)的读者认为保密性不重要。 语言AI,特别是机器翻译(MT)工具,正在达到更高的民主化水平。来自各种背景的人现在可以使用互联网网站和消费级应用程序执行多语言语音和文本转换和生成。 面向消费者的语言AI正在将自动字幕、机器配音、语音翻译和AI文本生成等技术引入桌面和移动设备,包括多语言通信和Web插件。这些消费者并不太关心最终质量,他们大多试图在人工智能允许的范围内尽可能用其他语言进行交流,通常会明确表示是人工智能翻译了文本或语音。 但在消费者可以轻松使用语言AI之前,个别内容创作者已经在以商业方式进行尝试,使他们能够在视频平台和网站上接触到更广泛的受众。像自动转录服务、字幕和标题这样的东西越来越便宜和容易获得。许多也是免费的,达到一定的数量。 语言AI正在以前所未有的水平为每个人扩展本地化能力,因此我们请Slator读者分享他们的朋友和熟人(语言行业以外)是否使用语言AI技术。超过三分之一(35.4%)的受访者认为有时会使用AI,另外三分之一(31.3%)经常使用AI。他们还发现,大约四分之一的人可能尝试过(27.1%),其余的人(6.2%)从未尝试过。 Slator在2024年7月对人工智能与人工翻译主题的文章进行了分析,强调了一些反复出现的主题,大约四分之一的抽样出版物提到人工翻译是“不可或缺的”。 随着人工智能翻译的普及,人们担心在创造性的、文化上准确的翻译中会失去人情味,很大一部分媒体报道赞扬了人类翻译的能力,并将其与人工智能的明显局限性(如幻觉和使用不正确的术语)进行了对比。 更大部分的报道(40%)将翻译人员生计的下降描绘为不可避免的事情,理由是人工智能导致的失业和裁员。报道倾向于使用一些熟悉的限定词,比如对人工翻译的需求下降,人工智能对翻译工作的威胁,甚至是对外语教育终结的猜测。 我们询问读者,他们认为主要媒体与人工智能相关的报道有多准确,超过三分之一(36.8%)的人回答说这不是很准确。两组人几乎平均分为认为它有点准确(23.7%)或非常准确(21.1%),其余的人认为它一直是错误的(18.4%)。

以上中文文本为机器翻译,存在不同程度偏差和错误,请理解并参考英文原文阅读。

阅读原文