Artificial Intelligence & The Human Factor
Artificial Intelligence & The Human Factor
- By Jorge Macias
- November 11, 2020
- 4:50 pm
Technology has made astounding strides in recent years, revolutionizing the translation industry from being solely based on human capital to relying significantly on the assistance of machine learning and artificial intelligence (AI). With the internet’s widespread use, more languages have become accessible, providing a global vocabulary database for sophisticated platforms to draw from. Although these advancements in translating methods have had an impact, the presence of human touch is undeniably necessary, even if its role has changed. The translation process is very telling in this regard, with three key elements touching on both the necessity of human editing and the swiftness of neural machine translations (NMT).
Confidentiality – Sensitive information is inherently encountered when handling financial or legal documentation for SMBs, large corporations, and non-profit organizations. Data breaches have unfortunately become commonplace, growing at an exponential rate and requiring language service providers (LSPs) to rely on cutting-edge technology to keep client information strictly confidential. As such, quality translation management systems (TMSs) allow for limited or partial accessibility for translators working on a highly sensitive document. Instead of having complete access at all times to the translated text, the TMS segments the document into various files, prohibiting the translator from viewing the original document and ensuring privacy is protected.
In a time when the travel and tourism industries have suffered greatly due to COVID-19, these easily avoidable blunders are too costly to take a machine-translated gamble on. The lack of human touch, nuance, and localization challenges make the convenient benefits of Google translate largely insignificant in the long term – pointing to the irreplaceable nature of language service providers.
Although advancements in translating methods have had an impact, the presence of human touch is undeniably necessary, even if its role has changed.
Translation Memories – Much of what makes NMT an attractive option is its affordability and speed. In combination with a comprehensive translation policy set in place by an LSP, neuromachine translations can easily pick up on repeated phrases, words, or terminology, subsequently creating a database of translation memories. These memories eliminate the need for repeated translations, greatly reducing turn-around time and costs to the customer, especially when charged per word.
Terminology – Part of any respectable LSP’s approach to translation is understanding the nuances of the industry a client operates in. AI and machine learning can assist in this endeavor but fall short when considering contextual clues or fitting the tone of a particular audience. A good TMS can be fully put to use by creating a glossary of internal terms that a specific client uses and locking in such terms so as to avoid unintentional tampering by translators or any other individuals. This is critical to assisting translators working on a particular client, since many times a corporation may have an internal language that competitors or other companies within the same industry would hardly even recognize.
The common thread witnessed in the successful implementation of a TMS is a proven protocol that translators can easily follow and rely upon when encountering the more complex, nuanced aspects of translating. Sensitive information, repeated phrases, and company-specific lingo can all be handled smoothly and intelligently when combined with the experiential knowledge of translators who are adequately trained to fill in the less-obvious blanks.
