Click on the 'Listen' button above to hear this interview.

Several high school sex ed classes around the country have traded in the quiet, lifeless sack of flour for robot babies.

In 1993, a former NASA engineer decided flour babies weren't a good substitute for the real thing, so he developed an electronic baby doll that would cry at random. To keep the baby "alive," teen caregivers need to preform an action to sooth the digital infant. These robo-babies are supposed to give students an idea of how difficult it is to be parent and help prevent teen pregnancy.

But new research recently published in The Lancet showed the opposite: In a study following thousands of teenage girls in Western Australia, it turned out the girls who took care of the robot baby were more likely to get pregnant before they turned 20. It was a small difference — 17 percent of the girls who got the babies became pregnant, compared to 11 percent who simply got a sex ed class.

The study's author said the difference was small, but significant. Meanwhile, the CEO of RealityWorks, the company that makes the virtual babies, distanced the company from the results.

"The study... was not a representation of our curriculum and simulator learning modality but the researchers ‘adaptation’ and is consequently not reflective of our product nor its efficacy," Timm Boettcher said in a statement.

Nicole Cushman is the executive director of Answer, a program at Rutgers University that advocates comprehensive sex education — that means NOT abstinence-only — and helps schools implement these programs. She explains why these robo-babies aren't effective in preventing teen pregnancy, and what actually does work.