Last week, the Associated Press
announced that it will use “robot journalists” to cover NCAA sports, adding thousands more stories to the AP wire.It’s a task that seems impossible: how could a machine learn to write like a human? It turns out that it’s not that hard. Software like Wordsmith, a content generation tool that the AP will use for some of its college athletics reporting, digests data and analyzes it to extract interesting pieces of information, then enters what it’s selected into a formulaic, readable chunk of text. There are, in fact, many more complicated things that robots can’t yet do.
For instance, laundry.
Here’s Timothy Aeppel, writing for The Wall Street Journal:
No machine can yet match a human’s dexterity and problem-solving abilities when attacking a pile of irregular shaped clothes of different fabric types and weight. The difference between picking up a lace nightgown versus unraveling a pair of crumpled jeans knotted with other clothes is a calculation that requires massive computing power and a soft touch.
Over at Matter, Alexandra Ossola
writes that a robot programmed to do laundry faces 14 separate tasks that all come saddled with varying degrees of uncertainty ( read the Matter story to find out what they are). Most robots right now can only complete half of these tasks. Here’s Ossola:
Programmers and artificial intelligence experts have spent decades designing increasingly sophisticated robots that can navigate these unpredictable situations, tweaking the software and smoothing the physical movements so that the robot can behave more like a human when faced with uncertainty. Some robotics experts are designing robots with sensors that can learn every time they move. Others are literally teaching their creations to complete certain tasks by reading the instruction manual .
At MIT, scientists have built a robot arm with spongy skin that can apply the right amount of pressure necessary to pick up different kinds of fabrics. And other researchers are coming up with sophisticated algorithms for laundry-folding:
If scientists can figure out exactly what it will take for robots to become more laundry-savvy, they may be able to usher in a new wave of automated technologies. But even the AP’s “robot journalists” haven’t yet displaced any jobs. Rather, these writers are merely “filling in the gaps”: a form of complementarity that NOVA Next editor Tim De Chant reported on back in October:
Nearly every worker today is complemented by some form of automation, whether they realize it or not. Trading algorithms give stock brokers an upper hand, iPads help appliance salesmen field consumer’s questions, and engine computers help mechanics diagnose a car’s problems. It’s part of the reason why productivity increased during the Great Recession, even as people were being furloughed or laid off. Those who kept their jobs were able to be more productive, thanks in part to automation.
