Is The Financial Industry Ready For Machine Translation?

On March 1, 2022, the National Institute of Information and Communications Technology (NICT) and the Financial Services Agency of Japan (JFSA) worked together to enhance bilingual communication within Japan's financial sector. Their efforts led them to create an AI-based engine that efficiently translates financial documents from Japanese to English and vice-versa.

They claim that the machine translation (MT) engine has reached "the level of professional translators specialized for the financial sector." Is this even possible?

Today, we will discuss the implications of this for financial translation services and how it will affect the international financial sector.    

  Human Parity in MT: Is It Possible?

Many MT specialists believe that one day MT will achieve the same skill as linguists. Human parity means that the MT engine's translation is equivalent to human translations.    

Besides the JFSA and NICT, many corporations and organizations in the past have claimed parity for their MT engines. However, after some evaluation, the MT engines still showed inaccuracies.    

Despite having a projected CAGR of 11% from 2020 to 2028, the MT sector is still young. Even with the rise of neural mapping and AI in translation applications, the accuracy of translations is still not at the same level as human translators.  

Based on what little information we could gather about the AI-based engine that the JFSA and NICT created, researchers found that translations deemed NG (No Good) were reduced to 12% from 21%.         

This is impressive, especially since financial translations are highly specialized needing expertise in the field. This new development in MT is a new hallmark toward human parity in highly specialized fields, such as the global financial sector.

  Where Will the Financial Translation Sector Go From Here?

Even though MT engines have come a long way since the 1950s, it's still not a hundred percent accurate. According to the same report by JFSA and NICT, when compared to other general-purpose MT engines, the percentage of human parity at the same level as financial translators was from 21% to 49%, which is impressive but not enough.

There's a reason why many individuals still seek professional translators even though MT engines are a cheaper alternative. So far, no MT engine has absolutely achieved the same level as human translators. Although many translators use them and translation tools, the specialized forms of translations require expertise and translators that have studied in the field for several years.

For example, financial translators are required to understand the financial system of two or more countries. Even though the two markets use the same language, like English, the translated documents need to follow the same format and common financial terms where they will be used.

MT engines can efficiently translate documents word for word, but that doesn't mean that the translated text would make sense contextually. Because currently, MT engines can't properly distinguish the cultural and linguistic nuances required in highly specialized translations like financial translations, it still needs to be overseen by humans.

  Final Thoughts

Some believe that we will accomplish human parity by 2029. According to Multilingual, some MT enthusiasts have claimed that "singularity" will most likely occur and render humans obsolete.

But culture and language evolve alongside society, which leads to terminologies and phrases constantly changing in meaning over time. Due to this, it will most probably take JFSA and NICT's AI-based engine a couple more years before it can be on par with financial translators.

For the time being, we would still need linguists specializing in translating financial documents. However, with the new developments, they'll be aided by more specialized and advanced technology that will help them to create accurate and fast delivery of financial translations to the global market.

About the Author:

Ofer Tirosh is the CEO and founder of Tomedes, a leading translation agency that provides language and technology solutions to businesses and Fortune 500 companies worldwide.