Using AI to Bring Diversity to Our Communities
We’re excited to introduce some new capabilities we have developed that will soon be part of our Mayo Clinic Connect community for patients and caregivers, but that we’re applying starting today here in the Mayo Clinic Social Media Network.
We’re using machine translation and artificial intelligence (AI) to enable Spanish speakers to more fully participate in our communities.
From now on, visitors to MCSMN who have Spanish as their browser’s designated language preference will have the opportunity to translate discussion board comments from English to Spanish by clicking a link that says Ver Traducción (view translation).
They will be able to respond in Spanish, and then English speakers will be able to view the machine-translated English version of those comments by clicking a View Translation link.
This is part of a project we’ve undertaken mainly for our Mayo Clinic Connect community for reasons I’ll discuss in a bit, but we wanted to start here in MCSMN for several reasons:
- MCSMN is our learning laboratory. We like to try new features here first. For example, we implemented our new front-page design here in November, and plan to complete a similar overhaul on the front page of Mayo Clinic Connect in the next couple of months.
- As members of MCSMN, you are our fellow explorers of new technologies. That’s why you’ve joined: to stay up to date on the latest advances and to learn together with us. We know you’ll find it interesting to see how it works.
- We expect it will be a good feature, but it won’t be perfect from the start. Before we apply it in patient-to-patient communications, we want to work out the kinks.
We think this is going to provide significant value to our online communities as we seek to serve an international and diverse audience. In Mayo Clinic Connect, for example, we currently have 65 discussion groups: 64 that are focused on different diseases, conditions or roles, and one that is Español.
That’s not an optimal way to include Spanish speakers: having them channeled into one catch-all group means they can understand each other but they aren’t necessarily interested in the same medical topics.
And if we were to create a parallel scheme of 60+ groups in Spanish, we wouldn’t have the critical mass of users to sustain a conversation.
The best of both worlds
We believe that by implementing AI-powered user-controlled translation we will have the best of both worlds. People who share health concerns will be able to connect with others in the same situation, regardless of whether they speak the same language.
By making the translation user-controlled, we expect to maintain fast page-load times while also making it clear that it is a machine translation. In our early testing the translation quality has been quite good, and because this is a machine-learning process, it will get better over time.
Eventually we plan to offer additional languages once we have Spanish working well.
By rolling the capability out first here in the MCSMN community we’re testing first in a lower-stakes environment, with just one group instead of 65.
So here’s how you can help:
- Check out this thread that we started in Spanish and see how the functionality works.
- If you have colleagues who are Spanish speakers or bilingual, please invite them to join MCSMN and to participate in the discussions.
What do you think?