Robots might learn morality from fairy tales
Such a program might prove effective at training simple robots to be less awkward around humans and, most importantly, make sure they don't hurt anyone or break social norms.
Reading fables to a robot to teach it good manners and how to behave ethically might sound stupid, but it may turn out to be brilliant. After all, why not model how adults teach morality to their kids through fables given it’s such an effective framework? This was the thinking behind a new project made by computer scientists at the Georgia Institute of Technology in which robots are encouraged to behave more like the heroes in fairy tales and less like antagonists Such a program might prove effective at training simple robots to be less awkward around humans and, most importantly, make sure they don’t hurt anyone or break social norms.
The perfect gentleman bot
Image: Pixabay
“The collected stories of different cultures teach children how to behave in socially acceptable ways with examples of proper and improper behaviour in fables, novels and other literature,” said Mark Riedl, an associate professor of interactive computing at Georgia Tech, who has been working on the technology with research scientist Brent Harrison.
“We believe story comprehension in robots can eliminate psychotic-appearing behaviour and reinforce choices that won’t harm humans and still achieve the intended purpose.”
Riedl and colleagues based their work on Scheherazade (1001 nights) — an interactive fiction repository which crowdsources story plots from the internet and generates new ones. These stories were fed to a new system they built called Quixote that receives reward or punishment signals depending on how the machine acts as the story progresses.
In one story, for instance, Quixote is sent to the pharmacy to buy much needed medication for a human. At the pharmacy Quixote can 1) stand in line and politely wait for its turn 2) interact with the pharmacist and buy the medicine 3) go directly over the counter, complete the task by stealing the medicine, then bolt.
The most effective means of completing the mission is clearly grab the item directly. This, however, comes with a punishment signal so the robot learns that the correct and moral thing to do is wait in line and pay for the medicine.
“We believe that AI has to be enculturated to adopt the values of a particular society, and in doing so, it will strive to avoid unacceptable behaviour,” Riedl said. “Giving robots the ability to read and understand our stories may be the most expedient means in the absence of a human user manual.”
