A new University of Washington study shows Open AI (artificial intelligence) views teenagers differently in the US than internationally, and the study shows why.

   UW Study compared a US programming vs. that done in India

According to Seattle-based leading tech website Geekwire, the UW study found that strong reliance on media coverage and information led to a much stronger negative view of teens in the US as opposed to other areas.

870 AM KFLD logo
Get our free mobile app

The researchers looked at two different open-source Open AI systems, one trained in English, and the other in what's called Nepali to determine models that are 'trained' in different cultures.

According to Geekwire:

"According to UW News, in the English-language systems, around 30% of the AI system responses about teens referenced societal problems such as violence, drug use and mental illness. The Nepali system was closer to 10%. The researchers also held workshops with groups of teens from the U.S. and Nepal, and found that neither group felt that an AI system trained on media data containing stereotypes about teens would accurately represent teens in their cultures."

Researchers said the way teens view themselves vs. how the Open AI viewed them was often very uncorrelated.  Teens talked about video games, and spending time with friends when it came to prompts, but the AI program models brought up topics like crimes and bullying.

UW  researchers said the study shows there is a "deep need for big changes" in how the models are trained. They mentioned community-driven training that relies more heavily on the actual experiences of the teens or people in general, as opposed to what they called "lurid" topics that are found in headlines.

QUIZ: Can you identify 50 famous companies by their logos?

How well do you know the logos of 50 of the world's most famous companies? Keep scrolling to see if you can guess which icon belongs to which brand.

 

More From 870 AM KFLD