Evaluating the UX of AI-Powered Translation Experiences

A global technology company asked us:

How do we better support AI-powered translation software users and be more inclusive of lesser translated languages and dialects?

More than 7,000 languages are spoken around the world today, not including languages that have been lost over time. Because English is currently the most utilized language in business, professional translators and interpreters, as well as the general population, often turn to online translation tools for help. 

Our client wanted to better understand: How do translators define quality? What happens when the grammar is incorrect or the output is incomplete? What ethical considerations should be taken into account for online translation resources?

And, what happens when the software doesn’t include the language or dialect you need to translate, like Catalan, Kurdish, Urdu, or Moroccan Arabic?

We answered:

AnswerLab designed a study with a range of international participants who spoke one or more of 15 specified low-resource languages. A language is defined as low-resource when it lacks data that can be used for machine learning and training AI systems, in contrast with Chinese, English, or Spanish, which are considered high-resource languages.

The participants needed to be native speakers of their language with professional or academic experience translating their native language into English and vice versa.

We conducted 90 minute in-depth interviews with each participant to uncover insights into both practical and ethical considerations of AI translation services. 

Outcome:

Our team of UX researchers captured user insights and pain points regarding online translation platforms, while identifying universal support for translating lesser known languages and dialects. 

Through our research we also gained a deeper understanding of ethical issues and challenges when translating these languages through machine learning. 

Some of our recommendations included: 

• Create buy-in with local communities and organizations to aid in the translation support and software development
• Include both bilingual and and non-bilingual speakers for native language accuracy, and aim to include multiple dialects
• Increase accessibility across device types, platforms, and levels of tech literacy

Ultimately, by including more languages and improving the user experience through great design, AI translation can not only help the world communicate better, it can potentially preserve cultures and languages for years to come. 

...

Learn more about how our emerging technology experts can help improve your customers' experience and stay at the forefront of your industry.