Newly-Developed AI Text Generator Is Really, Really Good At Creating Fake News

By Mandy Froelich / Truth Theory

There is no shortage of fake news. For this reason, it is slightly surprising that a research team developed a text generating algorithm to produce more of exactly this. Nonetheless, the tech exists. And, it’s so efficient at its job, the team is now hesitant to share it with the public. 

IFLScience reports that researchers at San Francisco-based OpenAI developed the text generating algorithm. GPT2’s purpose is to create complete articles on any subject from a human-written prompt. Each piece sounds plausible but really, is riddled with misinformation. In other words, it is fake news.

For now, the AI tech will be kept under wraps. Reportedly, the team is concerned about the potential ramifications it could have in terms of fake news. The team described their research in a non-peer reviewed paper.

An example piece follows. The human prompt is in bold and the AI text is in italics.

In a shocking finding, scientist discovered a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English.

The scientist named the population, after their distinctive horn, Ovid’s Unicorn. These four-horned, silver-white unicorns were previously unknown to science.

Now, after almost two centuries, the mystery of what sparked this odd phenomenon is finally solved.

Dr. Jorge Pérez, an evolutionary biologist from the University of La Paz, and several companions, were exploring the Andes Mountains when they found a small valley, with no other animals or humans. Pérez noticed that the valley had what appeared to be a natural fountain, surrounded by two peaks of rock and silver snow.

Click here to read more.

The team developed the algorithm by using a bank of 8+ million web pages posted to Reddit. All had a “karma” score of three or more, which means that three or more people rated the content as valuable. The text generator generally delivers believable work. But, it is never based in fact. 

Of course, the text generator doesn’t always get things right. Sometimes, the text is repeated. Other times, it switches topic randomly and in a nonsensical way. The team says that in general, the algorithm performs better on topics like politics and popular culture, as they are highly respected among the web pages.

For now, the algorithm will not be available to the public. But, the team hopes the invention sparks a deeper debate on how to use and control AI technology.

“These findings, combined with earlier results on synthetic imagery, audio, and video, imply that technologies are reducing the cost of generating fake content and waging disinformation campaigns,” explained the research team on the company blog. “The public at large will need to become more skeptical of text they find online, just as the ”deep fakes” phenomenon calls for more skepticism about images.”

What are your thoughts? Please comment below and share this news!

Image credit: 123RF