#AskTheExperts - Automation and Business Model

#AskTheExperts - 自动化和商业模式

2021-03-30 17:00 GALA

本文共1383个字,阅读需14分钟

阅读模式 切换至中文

A new instalment of #AskTheExperts, a GALA blog series where we ask translation industry experts inside and outside the GALA community for their insights and advice on managing business processes and digital transformation. Do you have a burning or knotty question? Send it to us and we’ll ask our experts. Today we ask our experts to answer the following question: There are three different types of automation in our industry: linguistic automation, process automation and transaction automation. Linguistic automation includes everything related to machine translation, pattern detection, quality assurance, and speeds up the work of the linguist (hopefully). Process automation cuts the slack in company-internal processes, and speeds up the work of the project managers. Transaction automation connects different systems and enables seamless cooperation between two players, increasing transparency and standardization for everyone involved. Through automation the integrity of the workflow can be regained, points of importance harmonized, and this enables a more seamless workflow between end-customer and linguists, without damaging the interests of the LSPs. Web-based systems also lend themselves better to measuring the time spent on a task than desktop-based systems. Technology improvements enable a more requirement-compliant workflow for the end users, and honest time reporting fosters time-based models, where everyone is compensated according to the real effort, and it's also in the interest of the end-customer to provide a better experience for the linguists. With automation, we are making processes and people more efficient. Yet successful automation relies on establishing pre-defined sets of rules that produce consistent, repeatable processes. This requires evaluation and standardization of processes to ensure that automation can be leveraged to optimal effect. Over time we expect that AI will help us determine the rules based on data trends and patterns of behavior to speed up implementation and also improve success rates. The other big area for change is around linguist expectations. How do they fit into the business model and what do they have to gain? Automation improves efficiency within services management (i.e., giving more time back to the linguists to complete the work), and AI-driven products can improve translator productivity so they have more time on their hands. However, these improvements shouldn’t be seen as a ‘pay cut’ - instead, they leave room for translators to do higher value tasks, like more robust QA checks, constructive feedback loops for continuous improvement, improving language assets, and more. As well, they’ll then have more opportunities for work moving forward. To me, it all begins with the client’s needs—how they think of making their business or activity global or international or multilingual or multicultural. Any “push for automation” or “change in the business model” will be the result of the process that the client uses to achieve this. Language operations never exist in a vacuum; they are always part of how a product or service is made. Language operations are also never the end: multilingual or multicultural products/services are. To quote the late Theodore Levitt, “Customers don’t want a quarter-inch drill. They want a quarter-inch hole.” In the life of many clients, language operations (translation, localization) are an afterthought; they try to “globalize” the already finished product. Others, however, have begun to recognize that their products or services will need to be multilingual or multicultural by design, and they need to deal with language and culture as early in the product life cycle as possible. The more language operations become integrated into production processes, the less it will be possible to just sell translations by the word. Language services will become more complex because they require an understanding of the production process—and not because of the push for automation. In this constellation, automation is a by-product, a consequence, and not the trend itself. With the numerous benefits automation brings, it’s no surprise that companies are adopting automation at an accelerated rate. In the language industry, machine translation (MT) is already at the forefront of automated translation production. Our 2020 data shows that 56% of translation jobs had MT enabled which means that machine translation post-editing has become the dominant translation method in enterprise localization. Businesses, from tech providers to translation buyers, are investing in this technology. But how will this change language industry business models? MT will bring higher translation quality for a lower price which will open up new use cases for companies to introduce MT-based localization to new content types. For the TMS market, we could see a stronger focus on integrability to address the needs in specific verticals, a deeper MT integration, and a focus on MT post-editing instead of traditional translation feature development. Thanks to automation, translation services are becoming an always ready utility that can be easily integrated into existing operating models. New technologies available in the language industry enable companies to better integrate their existing content systems with translation management systems to automate their entire localization process, saving everyone in the supply chain time and effort, transforming the way businesses operate. Clients, for example, will start to see translation costs reduce as project management overheads decrease, and translators’ work days will be less labor intensive and less unpredictable as many of their administrative tasks (such as quoting and invoicing, submitting translated files back etc.) will automatically take place. For example, new cloud-based platforms combine Linguistic AI with advanced workflows to automatically route tasks to the most suitable translators. The content is then instantly available within their CAT tool and invoices are automatically generated once the work is complete. The whole process is streamlined from start to finish, decisions are data-driven and manual steps are minimized, freeing up time to ensure translations are completed and delivered on-time. Ultimately, it’s fair to say the push to automation will increase business efficiency and decrease operating costs for the language industry. Brands are powered by words. From advertisements and marketing websites to application user experiences on your phone or browser, the sheer scale of content is increasing, and the specific words that are used establish differentiation from one brand to the next. For the past decade, automation has already disrupted the business model for the language translation industry because it has eliminated the need for costly project management throughout the content lifecycle for both translation buyers and suppliers. The result is that translation buyers can launch their content faster with fewer resources. While automation makes scaling content a bit more palatable by eliminating the human effort surrounding the translation process, the industry is poised for change once again. There will always be a place for human translators, however machine learning systems including neural machine translation will play a bigger role; and, over time they will learn from their mistakes and improve. Pricing models will evolve and brands will be looking for ways to optimize their translation spend while maintaining a high-quality output. Having the right technology in place to produce words and monitor the effectiveness of the supply chain is key to how this future will be realized. Automation is a way to respond to new market needs. And they are: deliver more, faster and at lower unit cost. It does not mean that automation will eliminate humans! Well the contrary. In order to deliver more, faster and cheaper we need both - humans and machines/automation mechanisms. What changes is the role of humans and their position in the process. Since the market requires fast delivery - counting from the start of the process - a lot of preparation needs to be done upfront, by humans, in order to prepare the automated process well before the project starts. That means we need more experts who can configure processes and workflows and define possible automations in the flow of the process as well as in the execution of particular steps. To give an example: automating a workflow requires setting up the rules according to which the workflow automation system triggers the acceptance of one step and launches the execution of the next one, when moving the content along the whole “production chain”. On the other hand, automating a particular step, like translation as an example, requires training of machine translation engines, so that they can be applied within a predefined process.
GALA博客系列#AskTheExperts的新篇章,我们向GALA社区内外的翻译行业专家询问他们对管理业务流程和数字化转型的见解和建议。你有一个迫切或棘手的问题吗?请发给我们,我们将询问我们的专家。 今天我们会邀请我们的专家回答如下问题。 在我们的行业中有三种不同类型的自动化:语言学自动化、流程自动化和交易自动化。语言学自动化包括与机器翻译、模式检测、质量保证有关的一切,并加快语言学家的工作(希望如此)。流程自动化减少了公司内部流程的松懈,并加快了项目经理的工作。交易自动化将不同的系统连接起来,使两个参与者之间实现无缝合作,为每个参与者增加透明度和标准化。通过自动化,工作流程的完整性可以得到恢复,重要性得到协调,这使得终端客户和语言学家之间的工作流程更加顺畅,而不会损害语言服务供应商的利益。与基于桌面的系统相比,基于网络的系统也更适合于衡量在一项任务上花费的时间。技术的改进使最终用户的工作流程更符合要求,诚实的时间报告促进了基于时间的模式,每个人都根据真正的努力得到补偿,而且为语言学家提供更好的体验也符合最终客户的利益。 通过自动化,我们正在使流程和人员更加高效。然而,成功的自动化有赖于建立预先定义的规则集,产生一致的、可重复的过程。这需要对流程进行评估和标准化,以确保自动化能够被利用到最佳效果。随着时间的推移,我们期望人工智能将帮助我们根据数据趋势和行为模式来确定规则,以加快实施速度,同时提高成功率。另一个大的变化领域是围绕语言学家的期望。他们如何融入商业模式,他们有什么收获?自动化提高了服务管理的效率(即把更多的时间还给语言学家来完成工作),而人工智能驱动的产品可以提高译员的生产力,使他们有更多的时间。然而,这些改进不应该被视为 "减薪"--相反,它们为译员留下了空间,让他们做更高价值的工作,如更强大的质量保证检查、持续改进的建设性反馈循环、改进语言资产等等。同样,他们也会有更多的工作机会向前推进。 对我来说,这一切都始于客户的需求--他们如何考虑使他们的业务或活动全球化或国际化或多语言或多文化。任何 "推动自动化 "或 "改变商业模式 "都将是客户用来实现这一目标的过程的结果。语言操作从来都不是存在于真空中;它们总是产品或服务制造方式的一部分。语言运作也从来不是目的:多语言或多文化产品/服务才是目的。引用已故西奥多-莱维特的话,"客户不需要四分之一英寸的钻头。他们想要的是一个四分之一英寸的洞"。在许多客户的生活中,语言操作(翻译、本地化)是一个事后的想法;他们试图将已经完成的产品 "全球化"。然而,其他客户已经开始认识到,他们的产品或服务在设计上需要是多语言或多文化的,他们需要在产品生命周期的早期处理语言和文化问题。 语言业务越是融入生产过程,就越不可能只是按字数出售翻译。语言服务将变得更加复杂,因为它们需要对生产过程的理解,而不是因为对自动化的推动。在这个星座中,自动化是一个副产品,一个结果,而不是趋势本身。 由于自动化带来的众多好处,公司正在加速采用自动化,这并不奇怪。在语言行业,机器翻译(MT)已经走在了自动化翻译生产的前列。我们2020年的数据显示,56%的翻译工作启用了MT,这意味着机器翻译的后期编辑已经成为企业本地化中的主导翻译方法。企业,从技术提供商到翻译买家,都在投资这项技术。但这将如何改变语言行业的商业模式?MT将以较低的价格带来更高的翻译质量,这将为企业开辟新的用例,将基于MT的本地化引入新的内容类型。对于TMS市场,我们可以看到更加注重可整合性,以解决特定垂直领域的需求,更深入的MT整合,以及注重MT后期编辑,而不是传统的翻译功能开发。 由于自动化,翻译服务正在成为一种随时准备好的实用工具,可以很容易地整合到现有的运营模式中。语言行业现有的新技术使公司能够更好地将其现有的内容系统与翻译管理系统相结合,实现整个本地化过程的自动化,为供应链中的每个人节省时间和精力,改变企业的运营方式。例如,客户将开始看到翻译成本的降低,因为项目管理的开销减少了,而译员的工作日将减少劳动强度和不可预测性,因为他们的许多行政工作(如报价和开票,提交翻译文件回等)将自动进行。例如,新的基于云的平台将语言学人工智能与先进的工作流程相结合,自动将任务分配给最合适的译员。然后,内容在他们的CAT工具中立即可用,一旦工作完成,就自动生成发票。整个过程从开始到结束都很精简,决策是由数据驱动的,人工步骤降到最低,腾出时间来确保翻译的完成和按时交付。最终,可以说,对自动化的推动将提高业务效率,降低语言行业的运营成本。 品牌是由文字驱动的。从广告和营销网站到手机或浏览器上的应用程序用户体验,内容的规模越来越大,所使用的特定词汇建立了一个品牌与另一个品牌的差异。在过去十年中,自动化已经颠覆了语言翻译行业的商业模式,因为它消除了翻译买家和供应商在整个内容生命周期中对昂贵的项目管理的需求。结果是,翻译买家可以用更少的资源更快地推出他们的内容。 虽然自动化通过消除围绕翻译过程的人力努力使内容的扩展更容易接受,但该行业已准备好再次变革。人工翻译永远有一席之地,然而包括神经机器翻译在内的机器学习系统将发挥更大的作用;而且,随着时间的推移,他们将从错误中学习并改进。定价模式将不断发展,品牌将寻找方法来优化他们的翻译支出,同时保持高质量的输出。拥有正确的技术来生产单词并监测供应链的有效性是如何实现这一未来的关键。 自动化是应对新的市场需求的一种方式。它们是:提供更多、更快、更低的单位成本。这并不意味着自动化将消除人类!相反,自动化将消除人类。恰恰相反。为了提供更多、更快、更便宜的服务,我们同时需要人类和机器/自动化机制。改变的是人类的角色和他们在过程中的地位。由于市场需要快速交付--从流程的开始算起--大量的准备工作需要由人类提前完成,以便在项目开始前做好自动化流程的准备。这意味着我们需要更多的专家,他们可以配置流程和工作流,定义流程中可能的自动化以及特定步骤的执行。举个例子:工作流程自动化需要设置规则,当内容沿着整个 "生产链 "移动时,工作流程自动化系统会根据这些规则触发对一个步骤的接受并启动下一个步骤的执行。另一方面,自动化一个特定的步骤,例如翻译,需要训练机器翻译引擎,以便它们可以在预定的过程中应用。

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