“This is the year that people really realized that you can build general-purpose robots.” | James Somers writing in The New Yorker

A Revolution in How Robots Learn

A future generation of robots will not be programmed to complete specific tasks. Instead, they will use A.I. to teach themselves.

Illustration by Tameem Sankari

Every time a video would go viral showing a humanoid robot jumping, dancing, or doing some other remarkable physical activity, there was a feeling among the general public that our wildest science-fiction dreams were just moments from coming to life. 

But inside the field of robotics itself, developments were happening far more slowly than in other similar technological fields, and straightforward practical applications, such as making a cup of coffee, remained out of reach. 

“A hopelessness hung over the whole enterprise,” James Somers writes, in this week’s issue. 

But that is changing—fast. “The last two years have been a dramatically steeper progress curve,” Carolina Parada, the leader of the robotics team at Google DeepMind, tells Somers. 

“This is the year that people really realized that you can build general-purpose robots.” 

The key development is that robots will no longer need to be programmed; instead they will learn through artificial intelligence, and then share that knowledge widely. 

• “Once one robot has learned how to tie shoes, all of them can do it,” Somers writes. 
• “Imagine copying and pasting not just a recipe for an omelette but the very act of making it.”

Read or listen to the story

=