Alexa, Welcome to the Real World of Personal Assistants
Why robots need to see and explore the world
MIT Technology Review recently published an article, based on a presentation by Rohit Prasad at MIT Technology Review’s AI conference. Prasad is the head scientist of the Alexa artificial-intelligence group at Amazon. The title of the article is “Alexa needs a robot body to escape the confines of today’s AI”. The subhead was equally eye-catching, “The man behind Amazon’s voice assistant says AI programs need to see and explore the world if they’re ever going to attain real understanding.”
This clearly got our attention. It also got the attention of a number of Misty followers, who forwarded it to us along with their own fun remarks and points of view.
In designing Misty there was never a question as to whether she’d have a body and whether she’d have the ability to move freely in space. Autonomous movement is core to Misty’s being and ability to take on meaningful assignments. When you start looking at the design of these new assistants in our lives, you can boil down to three points the rationale for giving them a body and the ability to move: Context and Understanding, Human Interaction, and Utility and Fitness for Purpose.
Context and Understanding
The point of context, specifically around language understanding, is the one that is highlighted in the article; by nature, language is complicated and can be ambiguous. This makes it hard for our new assistants and in his example, Alexa, to fully understand. Context is essential for language to be fully understood. Per Prasad, “The only way to make smart assistants really smart is to give it eyes and let it explore the world.”
When designing a personal robot, one whose use cases involve human interaction, a form factor that has human traits can create a more approachable experience and better the uptake and use. That said, it is key to consider the construct of the Uncanny Valley. This construct hypothesizes that objects appearing too human are likely to elicit a visceral response or cold, eerie feelings. And we have all seen examples of robots that have crossed the Valley. There is a balance to strike and, in this timeframe where we are working to overcome a fear of robots, it is particularly important.
Utility and Fitness for Purpose
A primary driver behind the design of a robot is its intended use case or fitness for purpose. There will be multiple form factors realized relative to robot design relative to the utility of the given robot. If the imagined use cases involve locomotion, it is simple. The robot needs to have a body and the ability to move. With Misty, if it’s got a floor, she’ll roam it. The use cases Misty is likely to fulfill will couple her ability to roam along with her other natural capabilities like collecting information, interacting with her humans, reporting on what she sees, entertaining etc.
As proposed by the article, “this could mean that Alexa will one day live inside something resembling a robot with eyes, limbs, and a way of moving around.”
Welcome to the real world of personal assistants, Alexa.
Here’s the original MIT article if you are interested.