We’re excited to bring you this piece on robot-learning on WIRED. It cites Alison Gopnik, a UC Berkeley professor and one of iOi’s champions. She delivered our inaugural lecture and spoke about child development. Here her work is taken a step further into applying those methods of child development to robots.
Robots, who are by their nature unnatural, must adapt to ‘life’ in our world. They get used to new environments and try out their motor skills, just like young children.
Babies and toddlers can’t do anything for themselves straight away, but they learn, almost innately. And those years are critically formative. What’s more, most of this early learning is done through play.
Despite this, even young children are more advanced than robots, who need strict controls, inputs and programming to navigate their environment.
However, as technologies develop – thanks to human advances – could robots improve thanks to learning and play in the same way as children?
UC Berkeley psychologist, Alison Gopnik, argues that robots need childhoods:
What you need is a kind of helpless, not-very-strong robot, and then have that turn into a system that is capable of actually going out in the world and doing things
This is a radical departure from traditional robot-learning, which involves teaching movements one by one, or by random trial and error.
Gopnik says that robots might benefit from some of the experimentation and adaptation that comes so naturally to children when they play.
Could we harness robots’ capacity and imbibe it with a sense of childlike curiosity?