Wake up Wor(l)d
I was able to attend more than three conferences over the last week. My brain is still recovering from the information overload but I’d definitely do that again. One of the topics discussed was voice technology. Since working on my final year project the field has been of interest. I never fully comprehended the complexity of the field until I dived into my project. This post is about what I’ve learned so far. Not the technical details but the conversations surrounding a technology that’s becoming more ingrained in our lives.
Fair warning I’m not an expert. I'm just a curious little bee buzzing around and absorbing the nectar of knowledge. Looking past the cheesiness of that statement, the point is to have a conversation and help people understand their world a little better.
A few weeks ago I was attempting to make dumplings from scratch. I was having a nightmare with the kneading. I called out to Google assistant hoping to help me figure out how to salvage the dough situation.
“Hey, Google. How do you knead dough?”
The response was a list of types of doors. I tried again but this time it didn’t understand what I said. I thought maybe my Indian accent is dominant when I say the word. Maybe I’m not enunciating clearly. Maybe it didn’t hear me. After several tries, I decided to touch my phone with all that sticky batter on my fingers and look up the solution. I didn’t think much of the incident until I attended the conferences. My experience addressed many of the pain points that people brought up at their talks.
Nowadays 20% of Google search is done through voice. I hadn’t realized how much of voice technology we have been using. With more people getting comfortable with voice technology, it's adoption is going to increase. For user confidence to sustain, the conversation they have needs to be contextual. So it's important that both the participants clearly understand each other.
Language plays a key role in this process of understanding and it’s not simple either. A major issue in the field of voice tech is lack of diversity. Going back to my anecdote about the dumpling incident, my Indian accent shouldn't have made a difference. If American English accents are the only accents it understands, we're denying this technology to millions. With around 6500 languages in the world, designing voice tech primarily in English is limiting. Not just in terms of access but the potential of the technology as well. Think about other languages and combinations of languages such as Hinglish, Spanglish, and so on. We also need to consider accessibility with people who speak in sign language. Yeah, sign language was something that caught me by surprise but it shouldn’t have.
As we are building the future for voice technology we need to be responsible. We need to be aware of the possibility of implementing an existing cycle of bias. We need to embrace diversity. Voice technology is powerful and has great potential. Think about the conversations we want to create. If you need motivation think about how much easier teaching your grandparents how to do stuff on their phone is going to be.