With so much hype around artificial intelligence, and so much news around the impact fake news has via Facebook, I’m surprised no one has put them together.
I came across a share from a friend which was basically a fake (Kurt Cobane suggesting Trump would be president in 1993), so I looked it up on Snopes.com and commented with that link. My friend followed up with the Simpsons one, rinse and repeat.
Could Facebook automate that?
To identify fake news you could feed an AI two sources, fact-checking sites (e.g. Snopes.com), and the news shared on Facebook.
NB: It would need to OCR text in posts/shares so that it could read text embedded in images.
When the AI matches a new post or share with fake news it could give the user the opportunity to cancel, or mark it as comedy / satire, or post anyway.
The news feed algorithm could then downgrade fake news, or mark posts as (probably) fake, or just hide them entirely.
The key is to catch people at the point of posting it, otherwise Facebook would have to downgrade people’s posts without warning and they run into (perceived) censorship issues.
If that worked, you could try using live news from more trusted sources (if they still exist) to use as a source for actual news instead (or as well as) fact-checking sites. That might provide a more real-time check, but it might also prevent people posting things that the ‘trusted’ sources haven’t seen yet.
It is probably a small step overall, but perhaps it could help?
Update – 16th Dec 2016
One month later Facebook has engaged with Snopes and others to fight fake news. Not as automated as I thought it would be, but the UI is fairly close to what I’d outlined.