Media outlets have recently reported that AI-powered software can translate text as well as expert human translators. One example of this is the article from The Verge titled, “Google’s AI translation system is approaching human-level accuracy.”
But are AI systems comparable to human translators, like some of these headlines imply? Not even close. Though present day AI translation systems are capable of interpreting large amounts of data, they still fail at many essential aspects of translation, and are very far from superseding humans.
Machine Translation (MT) is a branch of AI concerning the use of software to translate text from one language to another. This is basically a system that can be trained to recognize patterns in data, then transform the input data (e.g. a French sentence) into a desired output (e.g. a sentence translated to English). The main flaw in these types of systems is their lack of common sense — context and knowledge about the world. These types of systems will often translate the literal meaning of the words it has been given, rather than colloquial phrases that may be used.
Moving forward, it is important to be aware of the limitations of AI translation when comparing MT systems to human translators. While human evaluation is better, it requires a lot of investment. At TransTrust, we use expert level translators who have experience in your industry. Our translations are understandable, and use common sense to make sure your message is clear and has the intended meaning.
Resource: