Machine translation powered by artificial intelligence continues to evolve, but the human touch will remain – computers

Today is Translation Day, a day that marks the importance of technology in breaking down the boundaries of communication between people of different languages. THE artificial intelligence and machine learning algorithms are increasingly fine-tuned to offer spontaneous translations in real timebut above all to be more natural.

Although no one picks up a dictionary to try to understand words or text in a foreign language, but a smartphone resort to digital translators, and currently there is even equipment dedicated to automatic translation, such as Vasco V4, which supports 76 languages; it is believed that machines will never be able to perform translation 100% autonomously. Cultural issues, the age of the participants or even the context of the conversation are obstacles that machines should not be able to overcome and therefore the human factor will never be discarded.

IT IS it is precisely this symbiosis between artificial intelligence technology and human auditing that Portuguese scaleup Unbabel has been exploring for the past nine years. SAPO TEK had the opportunity to interview Vasco Pedro, CEO and co-founder of Unbabel, to better understand the current state of translation technology. THE the company’s platform allows instant translation of 95 language pairswith a mission to help create a world where language is no longer a barrier to communication between all cultures.

For Vasco Pedro, automatic translators play a fundamental role in the democratization of information. “It was precisely in this sense that they developed, in order to remove every obstacle standing on the way to a world without language barriers”. He believes that if we live in a global community today, it is because of the efforts made by the research field of machine translation or machine learning.

“It is precisely in this sense that automatic translators were developed, to remove every obstacle that stands on the way to a world without language barriers.”

He states that in recent years there has been an increasing demand for machine translation solutions. And that is due to the difficulty in contacting a professional human translator whenever something needs to be translated. Software is a suitable choice, but balancing the human being with artificial intelligence, “to get the best of both worlds”. As for the evolution of the sector, it is considered that it will continue to improve, with a balance, since “technology must be applied in areas where it is better than people”. And in the case of translators, humans must intervene when technology is unable to produce quality translations, especially in analyzing and interpreting the context and tone of the message.

not babel

Vasco Pedro, co-founder and CEO of Unbabel.

” data-title=”Unbabel – automatic translations based on artificial intelligence continue to evolve, but the human touch will remain – SAPO Tek”>

Vasco Pedro, co-founder and CEO of Unbabel.

As for the Unbabel platform, Vasco Pedro says it has eliminated language barriers through a platform that combines AI with human review. “Currently, our users have the ability to translate parts and content such as emails and FAQs into 96 language pairs.“. He explains that the purpose of the platform is to offer companies the possibility of expanding their business on a global scale, helping to communicate in the native language of the countries in which they are expanding, increasing relationships of trust with their clients. His solution enables the demolition of all barriers that are set up in connection with language differences.

For Vasco Pedro, in the case of translation, there will never be a complete replacement of the role of human workers. “The success of the interaction depends on the receiver’s ability to interpret the message. It is a very unpredictable factor with many variables such as age, culture, gender, among others“. These elements are difficult to teach an algorithm powered by AI. Therefore, in the most complex interactions, it is essential that there is a human interlocutor to review all messages and ensure that there are no misinterpretations.

“The success of the interaction depends on the recipient’s ability to interpret the message. It is a very unpredictable factor with many variables such as age, culture, gender, among others.”

But on the other hand, Unbabel believes that in simple interactions, such as requesting data, it is not justified to take the reviewer’s time because the message is easily predictable. “It is now possible to validate sensitive data without assigning an agent to this task”.

Is it possible to achieve a more natural artificial intelligence and keep translations in context?

According to Vasco Pedro, if we consider that artificial intelligence lacks the ability to correctly interpret messages in accordance with the presented context, the solution involves investing in responsible artificial intelligence, adding that this is a topic that Unbabel has addressed. There is also an application by a consortium of companies for funds from the Recovery and Resilience Plan, which aims to make Portugal a center for the development of this type of solution.

We believe that it is crucial that artificial intelligence acquires, in a responsible way, the ability to understand the context in which it is inserted, without harming the interlocutors“. And the CEO of Unbabel leaves a real example of the case of medicine or justice, where translation errors can occur due to over-reliance on artificial intelligence that is not ready to deal with technical terms in different languages. And these examples also serve to show consequences that can be “extremely negative, and it is precisely for this reason that we insist so much on the development of responsible artificial intelligence.”capable of any context or occasion”.

The translation sector, in which Unbabel operates, was born out of the human desire to remove what is considered one of the barriers that separate the globalized world, language. “If we analyze the development of the sector, we see that technology is responsible for the rapid development of the availability of translation tools“. At the beginning of the Internet, a tool that was one of the main contributors to the democratization of information, all content was mostly written in English, so it was only available to a small part of the world, emphasizes Vasco Pedro.

“I believe this is a natural evolution that will continue to progress so that one day translation will be accessible to everyone, including those with little technological literacy”. and is given an example of how search engines already have pre-installed applications that allow you to translate the content of an entire website into any language with one click. “The impact that this tool has on the availability of content is immeasurable, it does not require high technological literacy, even if the translation is not 100% faithful to the original language”. However, he admits that there is still a lot to be done in this area, but he points out that he is on the right track to create a world without language barriers that separate the global community.

“I believe this is a natural evolution that will continue to progress so that one day translation will be accessible to everyone, including those with little technological literacy.”

Recently, Unbabel shared the Global Trends in Marketing Localization 2023 report, noting that marketers already want to create content internationally, and thus the need to translate it into different languages. Vasco Pedro says that this need has always existed, but that it has been intensified by the consequences of the pandemic. Is that after a period of uncertainty, the world is opening its doors again, which means an opportunity for companies to anticipate internationalization strategies.

In this sense, marketing teams are preparing for this global growth and anticipating possible language barriers. on joint study says 84% ​​of professionals invested in translation to increase income during pandemic, among clients, partners or employees. He believes that everyone appreciates the effort to communicate in their mother tongue, and English is only a short-term solution.

In conclusion, Unbabel models work efficiently thanks to investment in balancing technology and human review. Models are programmed to distinguish between social, cultural and other contexts, even though they cannot make that judgment themselves. Nicknamed “Unbablers,” the company’s employees are responsible for verifying the interpretation made by the AI ​​and giving the final message a “human touch.”

Leave a Reply

Your email address will not be published. Required fields are marked *