How not to motivate your voice assistant
Speaking of this year’s Samsung Developer Conference, Samsung’s head of software and services InJong Rhee once again tried to motivate Bixby by touting it as a replacement for touch interfaces. This is not a new line; when Samsung launched Bixby it did so by claiming that voice was a significant improvement over hard-to-use touch interfaces.
I have two issues with this claim:
- Claiming that touch interfaces must be hard-to-use because people only use 15% of the functionality of their phone daily doesn’t pass the giggle test. People only use 15% of the functionality of their phone daily because that’s all they need. Give people Facebook, messages, email, and a browser and they’re good most days. That doesn’t mean all the other apps and capabilities are hard-to-use, it means that people only need them in more specialized circumstances. I only rarely use tethering, but that doesn’t mean it’s useless or difficult to use; it just means I don’t need it that often (and when I do need it, I’m really glad I have it available). There’s lots of research showing the advantage of direct manipulation interfaces. Ignoring it makes you seem like an idiot.
- Proposing to replace touch interfaces, which tend to be pretty good at revealing their functionality, with a voice interface that gives you no clue what it can do is even worse. If you really believe that people don’t use the full functionality of their phones because they can’t figure out how to do so, why is giving them a voice interface that doesn’t reveal it’s capabilities an improvement? News flash: it’s not. Voice interfaces are worse at communicating their capabilities, not better.
And it gets worse. Voice interfaces can be more efficient than touch interfaces, but they need to be designed differently. You don’t design a voice interface to be equivalent to a touch interface (with the notable exception of designing for accessibility). If the interfaces support equivalent interactions, a well-designed touch interface will be faster. You design a voice interface to provide high-level shortcuts. Think about it: would you rather tell your phone “send a message to my wife that I’m on my way”, or “open messages, start a new message to my wife, enter I’m on my way, send the message”? The former is a high-level shortcut, the latter is touch equivalent. But Samsung seems to think that voice assistants need to be touch-equivalent (or “complete”, as the company terms it).
I’m hoping Samsung makes improvements with Bixby 2.0, but first they need to establish a better motivating premise.