So Fake

There's so much going on in the Fake sector that I just can't keep track; here's a few bits:

Author pulls software that used deep learning to virtually undress women
June 2019, Ars Technica

This is it. Deep learning, neural networks, generative adversarial networks, all that. We see now the world in the way it looked when the internet first started showing up in people's homes, but then all of the sudden, a phone could connect to the internet while you're walking down the street.

Deep learning is the robots doing things humans never thought a robot could do, and things we never even thought to ask for. It's absolute magic, and it's about to completely f--- our sh-- up.

Mona Lisa guest on TV? Researchers work out talking heads from photos, art
May 2019, phys.org

We can now extrapolate from photo to video. Is that what Mona Lisa really looked like? Does it really matter? It is astutely pointed out that unlike previous techniques, this one doesn't need 3-D modelling to make the leap.

And how?
"Lengthy meta-learning" that's how.

Facebook removes accounts from Russia, Iran for 'coordinated inauthentic behavior'
Mar 2019, Reuters

Has a nice ring to it. Reminds me of "Low-Credibility Information"

Anyway, the accounts were removed for their behavior and not their content. Let that be a lesson. Not sure if I see this as a good thing because it means that the screening algorithms are sophisticated enough to identify patterns of activity rather than simple vis/text content recognition, or it's a bad thing because it's getting harder and harder to get away with illicit activity on the web. Depends on which side you're on I guess.

Supporters in Trump Facebook adverts were actors
July 2019, BBC News

I think I'm not even mad.That's what stock footage is for, no? I mean, what is authenticity; Eiffel Tower in Las Vegas type stuff.

Melbourne fake Chinese beggars scam busted by police
July 2019, AUNews

Last but not least; it wouldn't be a report on the state of the fake without some of the old masters showing up.

From the same people who brought you the fake zoo with a legit dog purported to be a lion, we now bring you -- fake homeless people!

They are literally shipped from China to Australia to look disheveled and beg for money. I refuse to believe it but in the article they're talking about them clearing a few hundred dollars a day. Fake homeless people. It's one thing when you shave your head and throw on a saffron robe to be a begging Buddhist, but to be a part of an international underground beggar syndicate where you don't shower for weeks and lie prostrate on the ground, that's dedication.

*Actually, the more I look into this, the less funny it becomes. It's way more common than it should be, and devolves into maiming kids to use as bait; few people can resist giving money to a crippled kid on the street.


Post Script
On the Buddhist Beggar Syndicate and the geographic function of susceptibility:

"The men targeted out-of-towners, [Robert Hammond, executive director and co-founder of Friends of the High Line] said, adding that his office staff had a rule of thumb for watching the interactions: --Each second a visitor was willing to talk to one of the robed men was equal to 50 miles away from New York City that the person probably lived.-- New Yorkers would not give the men even a second’s worth of their time, Mr. Hammond added."
-The Fake Monks Are Back, Aggressively Begging
Christopher Mele, New York Times, July 1, 2016

Post Post Script

^Here we have a refined specimen. I can't tell you exactly what makes this jump out at me, but it screams "robot". Maybe it's a semibot, same difference, the purpose is to manipulate; there's something intentional about it. Maybe because people don't really comment on a message board in order to influence others, but to voice their opinion, and the two look pretty different. 

Notes
The spread of low-credibility content by social bots
Chengcheng Shao, Giovanni Luca Ciampaglia, Onur Varol, Kai-Cheng Yang, Alessandro Flammini, Filippo Menczer. Nature Communications. Volume 9, Article number: 4787 (2018)

Here we analyze 14 million messages spreading 400 thousand articles on Twitter during ten months in 2016 and 2017.