SoylentNews is people
SoylentNews
New research has shown just how bad AI is at dealing with online trolls.
Such systems struggle to automatically flag nudity and violence, don’t understand text well enough to shoot down fake news and aren’t effective at detecting abusive comments from trolls hiding behind their keyboards.
A group of researchers from Aalto University and the University of Padua found this out when they tested seven state-of-the-art models used to detect hate speech. All of them failed to recognize foul language when subtle changes were made, according to a paper [PDF] on arXiv.
Adversarial examples can be created automatically by using algorithms to misspell certain words, swap characters for numbers or add random spaces between words or attach innocuous words such as ‘love’ in sentences.
The models failed to pick up on adversarial examples and successfully evaded detection. These tricks wouldn’t fool humans, but machine learning models are easily blindsighted. They can’t readily adapt to new information beyond what’s been spoonfed to them during the training process.
(Score: 3, Insightful) by The Mighty Buzzard on Saturday September 01 2018, @02:20PM (7 children)
Abso-fucking-lutely. Why? Because someone has to be in charge of deciding what constitutes "lies, propaganda and hate" and that's power just fucking begging to be abused.
My rights don't end where your fear begins.
(Score: 0) by Anonymous Coward on Saturday September 01 2018, @07:01PM (6 children)
You have to treat 'incitement' and advocacy the same way. Speech is merely speech. Only one person is responsible for the decisions he makes.
(Score: 1) by khallow on Sunday September 02 2018, @02:29AM (5 children)
Not at all. If someone is organizing attacks or other violence via public communication (for example, the genocides in Rwanda were often directed via radio stations), that's not legitimate discourse.
(Score: 0) by Anonymous Coward on Sunday September 02 2018, @03:50PM (1 child)
Letting the government decide what is and is not legitimate discourse is dangerous. People are responsible for their own actions. If someone chooses to listen to someone preaching violence, then that is on the person who chose to listen.
(Score: 1) by khallow on Monday September 03 2018, @12:41AM
How about a jury of your peers?
(Score: 0) by Anonymous Coward on Monday September 03 2018, @01:11AM (2 children)
that's not legitimate discourse.
It's not for you to decide what is "legitimate discourse"... A person's decision to act violently is entirely personal, and only he is responsible. The participants are responsible for the attacks, not the "organizers"
via public communication
Oh, I see. It should all be done in secret, like the order to drop the bomb on Hiroshima
(Score: 1) by khallow on Monday September 03 2018, @06:27PM (1 child)
It's estimated that 900k people died as a result of the genocide and its coordination via public media like radio stations. I think I can make that judgment just fine.
But there's a big difference between having a bunch of people who are willing to act violently, and having those people act violently in a way that causes a lot of damage. Coordinated violence can be a lot more harmful than uncoordinated violence.
That raises the threshold on coordinated violence.
(Score: 0) by Anonymous Coward on Wednesday September 05 2018, @11:44PM
There is no logic in that whatsoever. You are merely playing a numbers game. The people who decide raise the sword are the only ones to blame for the bloodshed.