By Dan Rairigh (MATLM 2019)
The future of the language industry is bright. In a world where globalization brings us closer together, advances in technology make it easier than ever to communicate and conduct our work efficiently. The primary purpose of a machine is to facilitate a specific task; so, the question remains, why do so many of us fear the rise of artificial intelligence (AI)? Admittedly, the notion of a machine learning to navigate an area so intimately human as language is disquieting. Where do humans fit in an industry that is so eager to introduce machine learning technologies? It is human to be concerned. However, we do have a role to play in the age of AI; in fact, it will require effort on the part of humans if machine learning technologies are to assist us effectively. This essay seeks to address three essential steps that language industry professionals must take to maximize the efficiency of AI within our industry. First, we must accept the presence of AI – in its current state. Second, we must examine areas of our work from the perspective of automation; where can AI be implemented to make us more efficient? Third, we must actively invest in machine learning and contribute to its development. Acknowledging these needs, understanding the potential, and, most importantly, taking these steps will lay the foundation for a prosperous future for our profession.
AI-its’s here to stay
The hype surrounding AI is justified. We have been teaching (programming) machines since the 1930s, but even then, futurists could not anticipate where we would be less than a century later. The power and influence of modern AI is so impressive, that it evokes fear in the minds of many modern workers. As Priya Mohanty elucidates in her article, Do You Fear Artificial Intelligence Will Take Your Job? “There is valid concern that even as AI saves lives and helps businesses thrive, it will destroy livelihoods.” (Mohanty). Even as we fear the evolution of machine learning, we recognize that AI is certainly advantageous across various domains. Mohanty enumerates on the many unique uses of AI, but the fact remains that people still live in fear that their roles will become automated. Despite this valid state of unease, we must remember that AI helps to relieve us of some of our more tedious responsibilities. We can train machines to sift through data, extract strings, internationalize code, and oversee other traditionally time-intensive tasks. Moreover, AI already exists; it is not a looming threat in the distance.
Instead of ignoring its growing use in our industry (as well as countless others), we would do well to accept it as an aid to our current processes. The remaining skeptics can seek solace in the few areas where AI may still be inept at automating human skill. For example, let us imagine a scenario in which a company is collecting vendor feedback on a new translation management system (TMS). The company wants to improve their tool, but ideally in a way that also helps the company achieve quarterly goals for company growth and performance. Even a sophisticated AI tool will not necessarily know which vendor feedback to prioritize over the rest. In this case, a skilled vendor manager who thoroughly understands the goals of their company will, most likely, outperform an AI tool. Nonetheless, AI is here, and it’s not going anywhere.
Rising Star Scholarship 2019 Essay Daniel Rairigh
The Superhero you didn’t know you needed
Here we reach our second step: where can we implement AI within our current professional processes? As already mentioned, AI exists to make our work easier. It has its uses in some areas, while in others it remains frustratingly limited. Ideally implemented, human and machine can coexist and help one another to create new, more efficient processes. Indeed, with the adoption of AI, existing processes may require some re imagination. If we approach this step with an open mind, it has the potential to revolutionize the way we provide our services. So, where do we automate? This goes beyond the previously mentioned “tedious tasks” replacement. Consider a service that is genuinely difficult to provide within the language industry. Translating from Estonian to Mandarin Chinese, perhaps? For project managers, finding a human translator to work between rare language pairs is a painful quest – and even more painful when they pay for it. What if we could train a translation system to achieve the long-desired zero-shot translation (ZST)? Arle Lommel, localization guru and expert on the subject of machine translation, recently discussed Google’s recent strides toward ZST in his article, Zero- Shot Translation Is Both More and Less Important Than You Think. In his article, Arle emphasizes two significant achievements of Google’s trained engine: 1) It does not pivot, meaning that it does not first require translation to a more common language before providing the output translation, and 2) It is able to leverage data from multiple languages at a time, meaning that it can pull relevant data from any language that shares context with the subject matter and provide a more accurate output than MTs that use a single intermediary language (Lommel). Developments such as this demonstrate the enormous potential of machine learning and prove that it truly is an exciting time to work in our industry.
Humans: The future of AI
If we are to move forward as language industry professionals, we must all actively participate in the development of machine learning technologies. Machines will continue to learn as long as we continue to teach them; in this way, the role of humans is pivotal. This will require enormous amounts of data, especially when we are building translation systems. It is within our own interests to see AI technologies succeed, notably when they are built specifically with our industry in mind. Companies like Lilt are already doing this; in their case, they are matching the power of human translators with an interactive and adaptive tool that greatly improves the efficiency of their translators. As a result of embracing the machine learning evolution that so many still fear, Lilt is, “able to translate at nearly five times their normal speed while maintaining the same level of quality and accuracy” (Hinchliffe), while its, “carefully vetted translators contribute to the system’s continuous improvement and are chosen for their domain expertise and ability to localize with sensitivity to cultural nuance.” (Hinchliffe). In essence, they achieve two goals at once; they design their work around the coexistence of human translators and trained machines while simultaneously selecting translators who can, over time, improve the accuracy of their tool. Lilt is the perfect example of a company within our industry that is endeavoring to pair their human workforce with modern AI technologies. In order for our industry to grow, provide better services, and become more efficient, more companies must reassess their current association with machine learning technology and contribute to its development. Rising Star Scholarship 2019 Essay Daniel Rairigh
Meet your new friend
If there is one thing that is certain, it is that machine learning will have a lasting impact on the language industry. AI will not be leaving any time soon, and it is in our interest to bolster its development. By following the three steps outlined above, stakeholders in the industry will better appreciate the potential of machine learning within our profession and adequately prepare themselves and others to reap the numerous benefits that AI can offer. Revisit the first paragraph of this essay; specifically, the sentence which reads: “It is human to be concerned”. Humans may be concerned, but the computers we train are not. They are expertly trained to assist us with the tasks that we are incapable of performing, or otherwise too time-constrained to perform alone. Moving forward, we must view AI as our friend. Not to do so would negatively affect not only our immediate future, but our projected future as an industry. The concern will pass, and when it does, the machines will be waiting for us … to help them, help us.
Mohanty, Priya. “Do You Fear Artificial Intelligence Will Take Your Job?” Forbes, Forbes Magazine, 9 July 2018, www.forbes.com/sites/theyec/2018/07/06/do-you-fear-artificial-intelligence-will-takeyour- job/#399b7edf11aa.
Lommel, Arle. “Zero-Shot Translation Is Both More and Less Important Than You Think.” Common Sense Advisory, 8 Feb. 2017, www.commonsenseadvisory.com/Default.aspx?Contenttype=ArticleDetAD&tabID=63&Aid=37905& moduleId=390.
Hinchliffe, Tim. “In-Q-Tel Partners With AI-Human Translation Company Lilt.” The Sociable, 20 Dec. 2018, sociable.co/business/in-q-tel-ai-human-translation-lilt/.