How AI Tools Will Make Global Communication Better for Everyone

人工智能工具将如何让全球交流变得更好

2023-01-23 17:00 multilingual

本文共1215个字,阅读需13分钟

阅读模式 切换至中文

Generative AI tools have arrived, quite loudly, onto center stage, and in my opinion, this is not just part of another tech hype cycle. Rather, they deserve our eager applause. What excites me the most about these new shining stars capturing the attention of so many people lately is that I firmly believe they will make global communication better for everyone in the long run. Here are six reasons I think anyone working at, or for, any company that needs to communicate with people around the world should be optimistic about this particular moment of transition, and what it means for the broader content universe. Source quality is the lynchpin of enabling global communication. It’s something every localization professional wishes they could influence, in order to make global communication better for brands and businesses. For so many years, localization teams have dreamed of being able to get their hands on the source content to optimize it for localization. What we all wouldn’t give to make source content more understandable, more concise, less jargon-filled, free of grammar and spelling errors, and generally, more helpful to the reader. Why? Because when we take that content into 20 languages, we also multiply many other things by a factor of 20: the errors, the confusion, the ambiguity, the imprecision. In a world in which we have become far too tolerant of “garbage in, garbage out,” we finally have a chance to clean up and pre-process the source content, turning it into something we might actually feel deserves to be brought into 20 languages in the first place. Right now, many source content creators are experimenting with generative AI tools in order to optimize, refine, and publish more content in English than they ever thought possible. As they are doing so, they’re applying tools that clean up grammar and style, remove redundancy, and communicate more plainly and clearly than many writers do naturally, without an editing process in place. When companies create more content, this usually leads to (at some point or another), customers or internal teams in businesses requesting that a portion of this content be translated into other languages. While many AI tools are offering many languages, they are usually doing so via connectors to known machine translation engines. This means that translation is often happening automatically, at scale. Interestingly, as automated translation has become more widespread, many end users and customers are flagging their feedback that machine translation isn’t always sufficient to meet their needs. As such, even the strongest advocates of machine translation technologies have to start creating differentiated workflows in order to provide customers with the quality they really expect. Consider this. Given that we now live in a world in which any customer can get a “basic” translation for free using an online browser plug-in, their expectation is that companies wanting to sell to them will need to do a much better job than what they can get themselves out of the box. They will expect companies to raise the bar even higher, customizing translations with their own company-specific terminology and improving and enriching those auto-generated translations. Also, companies will start to get more savvy at measuring quality and listening to customer input, so they can apply machine translation where it makes sense, but customize further where their customers really demand a higher level of quality. One thing that thrills me is the chance for brands to stand out and be differentiated within the coming content tsunami by not only the quality of their content, but by their distinctive, unique brand voice. Content obviously needs more than just a strong, identifiable, and endearing brand voice to rise to the top. But let’s face it. Content that’s written in a signature style, by a talented writer, is always more enjoyable to read, no matter what it says. Content that brings people joy while entertaining or educating them about a topic tends to get shared more often. Content that gets more shares online gets more reach. Content that gets more reach is usually ranked higher by search algorithms and thus more traffic. And the virtuous cycle continues. This is why I believe one of the best investments companies can make right now is in style guides, glossaries, and other content-related assets in their source language, to ensure this style can cascade into other languages too, as they gear up to capitalize on the changes these AI tools will introduce over time. If we allow AI tools to help content creators along in their process, we free up the humans to do more additive, high-value work. While we know from experience in the localization profession that it can take decades for these types of technologies to become fully operationalized and reach their full potential, we can remain hopeful that these tools will ultimately be enablers of content quality as things play out over time. Indeed, I’m excited about the potential these tools have to help us to carry out what will soon be basic prompts like, “Rewrite this for an audience with a 2nd grade reading level,” or “Convert this text to a less formal style,” or “Remove any industry jargon from this paragraph,” or “Apply plain language best practices to this text.” The opportunities for customization are huge with these technologies, and in fact, they open up the possibility for a lot of personalization. Imagine custom style guides that can take your company’s own brand voice and style guidelines and convert any and all text into that style, in just a matter of seconds. As for the dreaded “copy and paste” commands we humans used to have to carry out every time we caught a capitalization error, or a branded term changed, may those rest in peace forever and be parts of our memory, so that future generations can do these things with one click, or a simple prompt that is one sentence long. The thing that all of us should be most eager to see happen isn’t just that more content will be available online, to more people, and that it will be higher quality, and available in more languages than ever. That’s great, but that isn’t the real value of this next wave of AI tools, in my opinion. The true value is in the speed. The acceleration factor is what we should all pay the most attention to. Consider the fact that many brands have had style guides and glossaries in place for decades. Many companies have had human editors for generations too. But never at any point in the past could the average person with internet access get access to a human editor who could clean up their content in real time, translate it into another language nearly instantly, and move messages back and forth between people in the blink of an eye. The acceleration factor that these content tools enable, more than anything else, is what I really believe will help us make progress and move us forward, making global communication better for everyone. The February issue of MultiLingual magazine will have a special focus on AI, MT, and the future of language technology. Subscribe now and receive the issue in your mailbox after it prints.
生成式人工智能工具已经相当大声地出现在舞台的中心,在我看来,这不仅仅是另一个技术炒作周期的一部分。相反,他们值得我们热烈的掌声。最近,这些新的闪亮明星吸引了这么多人的注意力,让我最兴奋的是,我坚信,从长远来看,它们将使每个人的全球交流变得更好。我认为,在任何需要与世界各地的人交流的公司工作的人,或者为任何公司工作的人,都应该对这一特殊的过渡时刻持乐观态度,以及这对更广泛的内容领域意味着什么。 源质量是实现全球通信的关键。这是每个本地化专业人士都希望他们能影响的事情,以便为品牌和企业提供更好的全球沟通。这么多年来,本地化团队一直梦想着能够掌握源内容,以对其进行本地化优化。我们所有人都不会给源内容更容易理解,更简洁,更少的行话,没有语法和拼写错误,并且通常对读者更有帮助。 为什么?因为当我们将内容转换为20种语言时,我们还将许多其他事物乘以20倍: 错误,混乱,歧义和不精确性。在一个我们对 “垃圾进,垃圾出” 过于宽容的世界里,我们终于有机会清理和预处理源内容,把它变成我们实际上认为应该被带入20种语言的东西。 目前,许多源内容创建者正在尝试使用生成式AI工具,以优化,完善和发布比他们想象的更多的英语内容。在这样做的过程中,他们正在应用工具来清理语法和风格,消除冗余,并比许多作家自然地更清晰,更清晰地交流,而没有适当的编辑过程。 当公司创建更多内容时,这通常会导致 (在某个时候或另一个时候) 客户或企业内部团队要求将这些内容的一部分翻译成其他语言。虽然许多人工智能工具提供多种语言,但它们通常是通过已知机器翻译引擎的连接器来实现的。这意味着翻译通常是按比例自动进行的。 有趣的是,随着自动翻译变得越来越普遍,许多最终用户和客户都在反馈机器翻译并不总是足以满足他们的需求。因此,即使是最强大的机器翻译技术倡导者也必须开始创建差异化的工作流程,以便为客户提供他们真正期望的质量。 考虑一下。鉴于我们现在生活在一个任何客户都可以使用在线浏览器插件免费获得 “基本” 翻译的世界中,他们的期望是,想要向他们销售产品的公司将需要做得比他们可以让自己开箱即用的更好。他们希望公司提高门槛,使用自己的公司专用术语定制翻译,并改进和丰富这些自动生成的翻译。此外,公司将开始在衡量质量和倾听客户意见方面变得更加精明,这样他们就可以在有意义的地方应用机器翻译,但在客户真正要求更高质量的地方进一步定制。 令我兴奋的一件事是,品牌有机会在即将到来的内容海啸中脱颖而出,并通过其内容的质量,而且通过其独特的品牌声音而与众不同。显然,内容不仅需要强大,可识别和讨人喜欢的品牌声音才能升至顶峰。但让我们面对现实吧。由才华横溢的作家以签名风格撰写的内容,无论怎么说,总是更有趣。 在娱乐或教育他们有关主题的同时给人们带来欢乐的内容往往会更频繁地分享。在线获得更多共享的内容将获得更多的覆盖范围。覆盖范围更大的内容通常会通过搜索算法排名更高,从而获得更多的流量。良性循环仍在继续。这就是为什么我相信公司现在可以做的最好的投资之一是在样式指南,词汇表和其他源语言中与内容相关的资产,以确保这种样式也可以级联到其他语言中,随着他们准备利用这些人工智能工具随着时间的推移将带来的变化。 如果我们允许人工智能工具在内容创作者的过程中帮助他们,我们就会释放人类来做更多附加的、高价值的工作。虽然我们从本地化行业的经验中知道,这些类型的技术可能需要几十年才能完全投入使用并发挥其全部潜力,但我们仍然希望这些工具最终将成为内容质量的推动者,因为随着时间的推移,事情会发展。 事实上,我对这些工具有潜力帮助我们执行即将成为基本提示的内容感到兴奋,比如 “为二年级阅读水平的观众重写”,或者 “将这篇文章转换成不太正式的风格”。或 “从本段中删除任何行业术语”,或 “将通俗易懂的最佳实践应用于本文。” 这些技术带来了巨大的定制机会,实际上,它们为许多个性化提供了可能性。想象一下自定义样式指南,它可以在几秒钟内采用您公司自己的品牌语音和样式指南,并将所有文本转换为该样式。至于可怕的 “复制和粘贴” 命令,我们人类过去每次发现大写错误或品牌术语改变时都必须执行,愿这些命令永远安息,成为我们记忆的一部分,这样子孙后代就可以一键完成这些事情,或者一个简单的提示,就是一个句子长。 我们所有人都最渴望看到发生的事情不仅仅是更多的内容将在网上提供给更多的人,而且它将比以往任何时候都更高质量,并且可以使用更多的语言。这很好,但在我看来,这并不是下一波人工智能工具的真正价值。真正的价值在于速度。加速因素是我们大家最应该关注的。 想想这样一个事实,许多品牌几十年来都有风格指南和词汇表。许多公司也有几代人的人工编辑。但是,在过去的任何时候,拥有互联网访问权限的普通人都无法访问人类编辑器,该编辑器可以实时清理其内容,几乎立即将其翻译成另一种语言,并在人与人之间来回移动消息眨眼。这些内容工具所带来的加速因素,最重要的是,我真正相信将帮助我们取得进步,推动我们前进,使全球交流对每个人都更好。 《多语言杂志》2月期将特别关注AI,MT和语言技术的未来。立即订阅并在打印后在邮箱中接收问题。

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