Original photo: Vin Cox, CC BY-SA 3.0 license. Photos b–f are derivatives of the original and licensed under CC BY-SA 4.0
People think they can spot fake images. They can't. So the fake news battlefield is about to shift from text to images.
From Axios Science editor Alison Snyder: People can detect a fake image of a real-world scene only 60% of the time, and even then can only tell what is wrong with the image 45% of the time, according to research published in the open access journal Cognitive Research: Principles and Implications.
Those findings are very different from those from a Pew study earlier this year, which found that an overwhelming majority of online users (84%) say they're confident in their ability to spot a fake news story.
Why it matters: As people get smarter about what fake news look like, fake news perpetrators are adapting their techniques from creating false news stories to manipulating real ones. This is often done with doctored headlines, images, videos and sound.
- Visuals: A Stanford University study shows how new technologies can alter facial expressions in real time to change the context of someone's reactions. Here's an example that a technologist did on George W. Bush's facial expressions during an interview last year — pretty crazy. The Economist recently profiled ways that people with little training or experience can doctor videos in a similar fashion.
- Audio: Axios' Ina Fried reported earlier this year that Canadian startup Lyrebird is touting technology that can mimic any voice from just 1 minute of recorded audio.
- Headlines: A conservative group backing a Virginia gubernatorial candidate altered the headline of a local newspaper to misrepresent the truth about an opposing candidate's position, causing the post to go viral.
Our thought bubble: If big internet platforms are having a tough time regulating fake news stories on their platform, imagine the chaos of trying to monitor doctored videos and photos …