Yes, AI translation is making leaps and bounds, and we're very excited about its potential! We, at Tarjimly, are always happy to hear about developments that further language access.
​

Currently, AI translation and interpretation for romance languages, such as French and Spanish, is quite good, but not perfect.
​

We expect AI translation to eventually get to a point to be relied one, but we are not there yet.
​

Machine translation could put some at ease if the topic is personal and not one they want to share with anyone else.
​

Unfortunately, like most things in life, development lags for the economically disadvantaged, and this applies to languages, as well. The development of AI translation for less economically powerful languages is delayed, resulting in common mistakes, and lack of clarity. It cannot be relied up for many of these languages.
​

Languages with less raw data available (e.g. books, scientific papers, emails, social media content, etc.) does not provide sufficient resources to train language-based AI on, leading to greater inaccuracy and more mistakes.
​

AI translation is not available for low resource languages, leaving many of the world's most vulnerable underserved. Low resource languages include but are not limited to languages with smaller populations, or those that are mainly oral and have few written resources.

AI and Machine translations often don't accomodate different dialects (e.g. Dari vs. Farsi Irani), leading to incorrect translations when non-standard vocabulary is used.
​

AI translation often does not understand stronger/less common accents.
​

For languages with colloquial dialects (i.e. spoken dialects that differ from written dialects), AI translation has not developed the ability to recognize and accommodate these differences in vocabulary, nomenclature, and structure. This often predominantly affects less educated populations who are less familiar with the standardized versions of their language.
​

AI translation does not always understand context, resulting in inaccurate translations.
​

Using AI translation in high stakes situations such as asylum applications, where consistency and small details are magnified, could negatively influence outcomes (e.g. rejection of asylum cases).
​

AI translation, currenlty, does not reflect tone of voice, which could be perceived as insensitive, make those sharing vulnerable stories feel unsafe.