In a discussion earlier this year among European media people, questions were raised about information and democracy:
Will artificial intelligence replace journalists? Does it endanger democracy? How can AI be regulated without curbing the innovations it might bring for the good of all?Jay Doubleyou: controlling ai - part one: the dangers of chatgpt
Similar questions were being asked at the time by UK media journalist Ian Burrell:
It is nearly eight years since the abuse of online personal information by data analytics firm Cambridge Analytica aided Donald Trump’s presidential campaign and spotlighted the future risks of high-volume voter manipulation. Since then, the giant social platforms have done too little to address the threat of mass misinformation, being preoccupied with recent economic headwinds that have momentarily stalled their previously inexorable profit growth.2024 is the year that AI will show its potential to influence elections
When it comes to imagery, misinformation can have an immediate effect, as looked by by Reuters this week:
Images of former U.S. President Donald Trump smiling as security personnel surround him during the assassination attempt Saturday are digitally altered. A Threads account, opens new tab shared an image of smiling security personnel as they surround Trump. The caption said, “Why are the secret service smiling? STAGED.”
Another person shared the altered image of Trump on X and wrote, “Trump is *smiling* AFTER surviving an assassination attempt! None of this passes the smell test.”
Both images were altered.
The original photographs, opens new tab of Trump was captured by Associated Press photographer Evan Vucci and does not show the former president smiling. It shows him looking forward with a serious expression while surrounded by security personnel.
Fact Check: Images of Trump, Secret Service smiling after shooting are fake | Reuters
Journalist Ian Burrell reports on how the AI is doing this:
"Fact-checker NewsGuard's analysis of 10 leading generative AI chatbots (including ChatGPT-4 and Microsoft's Copilot) found that they failed to provide correct information on the Trump shooting 57 percent of the time and supported numerous false narratives..."
PressReader.com - Digital Newspaper & Magazine Subscriptions
The fact-checker has a lot to offer:NewsGuard - Transparent Reliability Ratings for News and Information Sources
For example:
How Kremlin uses false fact checks to spread disinformation
NewsGuard launches suite of AI anti-misinfo tools | Semafor
Newsguard releases Israel-Hamas fact-checking resources | Media news
Newsguard downgrades credibility scores for NYT and GB News
41 TikTok accounts use AI to mass-produce political misinformation
The anti-establishment group 'New Tolerance Campaign' doesn't like them:
NewsGuard Must Answer for Fact-Check Bias - New Tolerance Campaign
This is a real issue, though:
Photo of Kamala Harris and Jeffrey Epstein Embracing on Beach is Fake | Snopes.com
AI image misinformation has surged, Google researchers find
There is an irony, however:
Finally, there are other fact-checkers:
But who can we trust?
List of fact-checking websites - Wikipedia
.
.
.
No comments:
Post a Comment