As if getting picked on at school wasn't bad enough, thanks to technology, kids and teens are now subject to the online torture known as cyberbullying.

Startled by the tragic stories in the news about suicides of bullied teens, and statistics that show the issue is growing, researchers at MIT have developed artificial intelligence to help manage the problem.

The Software Agents Group of the MIT Media Lab spent the last year or so working on its TimeOut project, which zeroes in on online bullying.

The goal isn’t to catch every instance of cyberbullying, but rather to prevent incidents from escalating.


"It targets possible cases of cyberbullying and also tries to provide targeted education for kids who are in those situations,” said Henry Lieberman, a principal research scientist at the MIT Media Lab and head of its Software Agents Group. “We are not trying to directly accuse anyone of being a bully, but we can use a situation as a teaching moment to educate people.”

Later this summer the group hopes to deploy software on MTV's, a forum for teenagers to post their problems anonymously and other teenagers leave comments giving advice.

The software analyzes the posts, many of which focus on bullying and concerns about sex, and uses an algorithm to identify certain groups of words within a post that could indicate possible harassment.

“Very often with bullying, kids don’t know whether situations are normal or they think it must be their fault somehow, and when kids internalize, that’s when you get problematic situations that lead to things like suicide,” Lieberman said.

The developers said they hope the software helps eventually lead to better discussion between victims, bullies, parents and social network moderators.

“We hope that it will have a positive impact. Of course you can’t force anybody to change, but we want to just try to provide educations for kids when these situations arise, and hope that their good sense will kick in,” Lieberman said.

Researchers looking at these possibilities to combat cyberbullying once it is identified:

  • Providing a link to educational material appropriate to the user’s situation. For potential cyberbullies, the material could encourage empathy for the victim and warn of possible damage to the bully’s social reputation. The intervention could encourage victims to get emotional support, learn how others have dealt with similar situations, give suggestions for appropriate responses (such as humor), and discourage them from retaliating. The material could also encourage friends to defend the victim rather than join in with the bully.

  • Instead of social network providers just offering a simple “Send” message, the button could be changed to remind the user of the consequences, and say, “Send to 350 people in your network.” Researchers suggest offering an “Are you sure?” button to pop up before a cyberbully sends a potentially problematic message.

  • Delivery could be delayed overnight to give the sender a chance to rescind the message in the morning before it’s actually delivered.