Artificial intelligence (AI) has been used to make major breakthroughs over the past few years in industries such as health care, cyber security, and logistics, giving rise to hopes and speculation about AI automated translations. AI has improved our daily lives with new automated technologies like when we search for a destination on Google Maps or choose a movie recommendation on Netflix. So why hasn’t AI been able to revolutionize the translation industry yet? The answer goes beyond technology.
What has been done?
Google Translate is the world’s most popular translation software which uses AI technology. It supports over 100 languages and serves over 500 million people every day. However, even the most advanced translation software in the world still experiences glitches, such as the infamous incidents in which Google translate started translating phrases into messages insinuating the end of the world and the second coming of Jesus. Although many netizens attributed the demonic messages to the works of Satan, it actually had to do with glitches in the machine learning which uses text from other sources on the internet to make sense of the inputted text.
Translation technology has come a long way in the past few years (as anyone who used Google Translate 5 years ago can attest). It is constantly being developed and many translation solutions are already using AI to improve their output. However, we are yet to find a technology solution that can create perfect human-quality translations.
Challenges of AI language processing
Despite recent advancements in natural language processing and sentiment analysis, many of the challenges of using AI to translate text have to do with the complexities of languages. Even though languages consist of sets of rules, principles, and processes, there are many nuances that cannot be detected or processed by machine learning. The reality is that AI still has problems processing requests in one language, without the additional complications of translation. The more languages a translation platform supports, the more learning it has to do and the more knowledge it has to store.
Another major challenge of teaching a machine to perform perfect human-like translations has to do with the difference between translation and localization. In order to fully localize content, one must be sensitive to cultural aspects, slang, and other regional language differences. Even with almost-perfect machine translations, it’s still crucial that content be written or reviewed by a native speaker with an in-depth knowledge of the local culture. Neglecting this may result in loss of investment or even a PR disaster.
Moreover, while utilizing an AI platform like Google Translate can provide quick and, depending on the subject matter and use, accurate translations, a large concern is data security and privacy, particularly for life science content. The implementation of new data privacy regulations in recent years, like the EU’s GDPR, has intensified the need for data secure technology across all industries, including translation.
Finding middle ground
The San Francisco-based startup, Stepes, has found a happy medium between using AI to automate and improve their app efficiency while ensuring that AI automated translations are still done by skilled linguists. The app gives users access to a cohort of freelance translators around the world available to translate content on-demand. Much like Uber, Stepes uses AI machine learning to improve their translator rating system and recommend translators based on their expertise and previous related projects. This way, projects are done using machine like speed and scale, but with human translation quality. As technology advances, hybrid solutions such as this are mostly likely to be successful.
Leave it to the pros
Despite advances in other industries thanks to AI, translation is still best left to the professionals. This is especially important for companies looking to expand to global markets. Many companies can’t achieve market penetration due to poor translation and localization and often end up losing their investment. In a quickly globalizing economy, word-for-word machine translations will not suffice. This is why it’s important to hire native linguists with cross-cultural communication expertise.
CSOFT has experience working with companies and organizations around the world to improve their content for global audiences. This includes technical translations, marketing and communications, and LT or app translations.
To learn more about translation and localization services offered by CSOFT, contact us here.
For more information on Stepes’ on-demand, AI automated translations, visit Stepes here.