As an academic, Cynthia Breazeal pioneered research into social interaction between humans and robots, developing Kismet, a robot that used facial expressions in a meaningful way. This week a company she founded announced Jibo, a prototype robot that aims to bring some of the ideas developed through this research to market. Jibo’s body and face are simple but emotionally expressive, and the robot responds to simple voice commands. For example, it can be told to shoot videos, relay messages, or operate “smart-home” gadgets. Breazeal spoke to Will Knight, MIT Technology Review’s news and analysis editor, about the inspiration for Jibo, and the challenges of commercializing social robots.
What’s so special about Jibo?
We’ve been creating social robots for many, many years. They’ve tended to be research robots—extravagant, expensive, temperamental robots. We can now create a social robot at a mass consumer price point.
When you think about how human beings really experience the world—think, act, and behave—it’s social, it’s emotional, it’s physical. A really simple example is that you can turn on the camera app on your smartphone, you can step out of the action, and you can take a picture. But a social robot plays that role on your behalf. You can ask Jibo to take a picture, and he goes into cameraman mode. But he’s also got autonomy, so he’s able to track faces and see where people are in the environment—so you can be in those pictures.
Verbal communication is notoriously difficult for machines. How sophisticated is Jibo’s voice interface?
One of the big lessons we’ve learned from a lot of AI is that the more you can scope the task, the better the chances that you’re going to deliver a good experience. So Jibo’s speech and natural language processing capabilities are purposely designed to support the skills that Jibo has.
Jibo is a charming, enchanting little persona, but he is really trying to support you specifically with what he can do. Of course, over time the skills will expand, but we’re not saying you can say whatever you want and Jibo will do something useful.
What have you learned since developing Kismet?
That people’s interactions with social robots combine the best attributes of human relationships, companion-animal relationships, and our relationships with technology.
We did this weight-management robot a number of years ago, and one of the fascinating things we found was that the physical attributes of the robot really tapped into people’s minds in a way that they were more engaged with the robot, and did better than when just using a computer, even when the information was identical. People also formed this emotional relationship with the robot that was almost more like [the relationship with] a companion animal. The human robot relation is something very special and very different. Once you understand that, you can start to leverage it in a way that really empowers people.
These weird virtual creatures evolve their bodies to solve problems
They show how intelligence and body plans are closely linked—and could unlock AI for robots.
A horrifying new AI app swaps women into porn videos with a click
Deepfake researchers have long feared the day this would arrive.
Chinese hackers disguised themselves as Iran to target Israel
But they left a few clues that gave them away.
DeepMind says it will release the structure of every protein known to science
The company has already used its protein-folding AI, AlphaFold, to generate structures for the human proteome, as well as yeast, fruit flies, mice, and more.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.