Artificial Intelligence in the translation market

November 24, 2023
Paula Ribeiro

Like in many other areas, the use of artificial intelligence (AI) in the linguistic sector has rocketed in recent years, because, and very simply said, AI is a much more affordable and faster option for organisations to create and locate their contents and reach their customers and consumers in a more agile and cheaper way, using new functionalities with which they are regularly “bombarded”.

However, while we continue to discuss the benefits of AI in the linguistic sector, could there be disadvantages that no one has addressed and issues that we should consider and take into account?

In contrast to the majority of automatic translation engines that can take months or even years to “learn” how to produce machine translation contents of good quality, one of the announced main benefits of AI in translation is the fact that it can generate translations immediately similar to those of humans but without any training [referred to as zero-shot learning] and produce enormous volumes of contents of acceptable quality in very little time, enabling faster entry into new markets.

Another advantage of using AI is that it is much cheaper than translation by human linguists, as human work is primarily focused on the revision and post-editing of the AI generated translations rather than on the actual translation. For organisations with smaller budgets and huge location searches, AI could be a more affordable way to locate contents in a timely manner without increasing their costs beyond that forecast.

Lastly, an enormous advantageous is perceived in the fact that human linguists can concentrate on tasks that add value to the process, such as ensuring that the content is properly located and uses the right pitch, or that it is compliant with the legislation of each market. In a nutshell, AI does the basic work and human linguists put the “finishing touches” so that the content can be published in a fraction of the time it would take were it only worked by human linguists along each step of the process.

But is it really that obvious and does it only show benefits? Let us see,

Language and culture go hand in hand. Making the contents resonate culturally is absolutely crucial for the brand’s image in its global markets. Although AI might actually do an excellent job of cultural location of contents in some markets (the most common), it might not have the necessary data standards to satisfy markets that are more restricted, more specialised, yet of no lesser importance. For this reason, it is essential to bring in a human linguist, as the publication of culturally inappropriate contents in a market could easily backlash on all the rest, especially in an era of social networks, where news circulates at lightning speed with a very wide-ranging reach.

Translation systems with AI frequently face challenges when involving specialised fields such as, for example, law or medicine, giving rise to translations that could be less precise or applicable in those specific areas. In areas where the accuracy of the content is so important, any mistake could be very expensive, which is why it might be faster for a human specialist to translate and/or revise the content rather than engage in a time-consuming post-editing task. Moreover, presenting an incorrect translation of such sensitive contents could severely damage the reputation, not to mention the profits, of a brand.

Translation by AI could process sensitive data, which raises questions about the potential privacy and security of those data. As learning models gather information from the Internet, personal or sensitive information could be used. The current lack of regulation means that the data can be shared without the data subject’s knowledge or acceptance. The topic of data protection and AI has not yet been fully resolved in the industry and still raises many questions and debates, one of the key issues being that it could have a major impact on the long-term viability of the use of AI to create and translate contents in certain industries.

The technology of translation with AI has clearly progressed enormously over the last few years, elevating the translation process to a much higher level of efficiency and speed. And while there are many benefits, there are also limitations. The bottom line is that it is possible to explore the opportunities and make the most of technology fed by AI in the manner most suited to needs and goals - and understanding its weak points will also help us to use AI in a more intelligent and effective manner.

[Image generated by Artificial Intelligence]

Paula Ribeiro
Latest posts by Paula Ribeiro (see all)