Google wants its robots to figure their s–t out.
The company’s artificial intelligence lab, DeepMind, is working to equip their AI with imaginations. This would allow them to make plans, anticipate consequences and, essentially, figure out how to achieve whatever goal or task they’re faced with.
“When placing a glass on the edge of a table, for example, we will likely pause to consider how stable it is and whether it might fall,” DeepMind wrote in a blog post about how humans use “imagination” to assess a situation.
“On the basis of that imagined consequence, we might readjust the glass to prevent it from falling and breaking. This form of deliberative reasoning is essential ‘imagination,’ it is a distinctly human ability and is a crucial tool in our everyday lives.”
The lab put the imaginative bots to test with a spaceship navigation game and Sokoban, a popular video game that requires the player to efficiently store boxes inside a warehouse – kind of like a simpler version of Tetris.
The AI “considerably” outperformed expectations, according to the blog post. It solved 85 percent of the levels it was tasked with, compared to 60 percent for AI that uses older models.
“They learn with less experience and are able to deal with the imperfections in modeling the environment,” the research team wrote. The robots were also better at learning different strategies and figuring out what they didn’t know.
DeepMind hasn’t cracked the code on imitating human intelligence but this is a step in the right direction if we want to live in a peaceful world where robots don’t crash cars, initiate anarchy or place a glass of water on a rickety table.