Two solutions to test for deep fake text
Two solutions to test for deep fake text
Synthetic content is probably going to be the gravest threat to the authority of media companies. Media newsrooms need to have a pro-active strategy to track, and flag deep fake text and video, which is now proliferating at an alarming rate.
Reassuringly, so are the solutions being developed to detect deep fake. Here are two that should be bookmarked by news teams.
Grover is an algorithm developed by researchers at University of Washington and Allen Institute of Artificial Intelligence. Fill in the text that you want checked, and it will detect whether it a neural fake news.
Another tool, developed by a team from Harvard and MIT-IBM Watson Lab, The Giant Language model Test Room (GLTR), uses its own predictive text output and compares the test article. It then checks if the top 10, 100 or 1,000 words of the predicted text matches the sample, and accordingly assigns a green, yellow or red rating.
Both GLTR and Grover are not solutions for deployment at scale, but they provide an early opportunity for newsrooms to adapt their existing production processes to handle the deep fakes in a pre-emptive fashion.
Image Source: Shutterstock