Azinno
ThinkTank-SignIn.png

AI Increases Participation Across Cultures

Using real-time AI translation to increase participation across language and cultures (2017).

ThinkTank-SignIn.png

About ThinkTank

ThinkTank is a cloud-based platform for leaders and teams to drive to amazing outcomes. This digital engagement systems helps companies companies cut through organizational politics and social bias, tear down silos, and reduce conflicts to align and engage distributed teams to execute faster.

The Challenge

Our challenge was to meet a board directive to protoype a new disruptive feature in 5 weeks. Based on deep UX field research and customer interviews over the previous 7 months, I suggested we break down language barriers that come from non-native speakers participating in live collaborative sessions, such as brainstorming.

 

Week 1 - UX Research

To kick off, the product manger and I conducted 1 week of UX research including interviews with users from around the world. A few quotes:

Chinese stop talking when language goes to English, they often don’t have the confidence to participate.
— Malay, Spanish, Norwegian session
When using ThinkTank, the nuances are very important.
— Swedish / English / Norwegian session
 
Lots of upfront translation is needed to support the sessions last year.
— Spanish / Portugese session
Been using Google translate to translate the results back to English.
— Portugese / English session
 

Research Insights

  • When using ThinkTank, the nuances are very important and quality is most important! The facilitator must trust the translation.

  • Global customers believed the concept would alleviate some of the cultural/ language barriers to result in better participation and outcomes.

  • If it works, the Return On Investment for the customer could be substantial.

Hypothesis

AI services in natural language processing, language translation, and dialogue could provide live translation across participant languages to get to better outcomes much faster. Possibly reducing total session time from 4 weeks down to 1 week. With my IBM Watson experience, I knew this could be a game changer!

Understand Translation Quality

With quality being the biggest potential show-stopper, over the next two days I leveraged my network of friends, family, and co-workers, who were native in other languages to test live translation services and APIs. The goal was to determine which service provided the best quality. Testing included Danish, Portuguese, Spanish, and Persian (Farsi), using Bing, Google, IBM Watson, and Yandex.

Google translate proved to be the best at understanding nuances and not translating words incorrectly.

 

Week 2 — My Concept Story

 
 
Nancy-Persona.png

Nancy is leading an implementation of a Human Resource Transformation project.

 
 Nancy has set-up the Globex Implementation project for her client.

Nancy has set-up the Globex Implementation project for her client.

 She makes the session visible to all participants.

She makes the session visible to all participants.

 Nancy adds a collaborative Crowdsource activity so everyone can contribute live from around the world.

Nancy adds a collaborative Crowdsource activity so everyone can contribute live from around the world.

 Later, as the facilitator, Nancy orchestrates the discussion in her native English language.

Later, as the facilitator, Nancy orchestrates the discussion in her native English language.

 
 
Jose-Persona.png

Simultaniously as a contributor, Jose is able to collaborate in Portuguese.

 
 Jose is experiencing all of the content translated live, as he adds comments to the discussion.

Jose is experiencing all of the content translated live, as he adds comments to the discussion.

 Jose can inspect the source text too.

Jose can inspect the source text too.

 

Weeks 3 to 5 — Coding and QA.

Customers loved the concept. Facilitators could save days off preparation, reporting, and leading to better outcomes. In the end, translation between two languages was feasible, but no translation service was robust enough to simultaniously localize between all of ThankTank’s supported languages. A few customers continue to run research projects with ThankTank to see how AI technology can improve their collective intelligence.