Next step: experiments in inducing thoughts of suicide in Facebook users.
The six simple questions Facebook refused to answer about its creepy suicide-detection AI
Facebook is using mysterious software that scours material on its social network to identify and offer help to people who sound potentially suicidal. It's some new technology CEO Mark Zuckerberg mentioned in his 6,000-word manifesto earlier this year. Yet, his lieutenants have kept quiet about how the borderline-creepy thing …
COMMENTS
-
-
-
Wednesday 29th November 2017 15:49 GMT bombastic bob
Re: I wonder how many false positives they're getting...
"From all those disappointed Trump-voters..."
not NEARLY as many as the number of REAL positives for Bernie voters who were screwed over by their own party (and especially Mrs. Clinton).
yeah you just had to go and try to slam on Trump, didn't you? (troll icon observed) [I'm a VERY satistfied Trump voter, thank-you-very-much, who's INFINITELY DIS-SATISFIED with what Con-Grab and "the Establishment" are doing, resisting the Trump plan, but i digress...]
but if the Faece-Bitch weenies are REALLY trying to manipulate people (and I'm sure they ARE), they'd generate 'False Positives' on those who ARE satisfied Trump voters that RABIDLY COMPLAIN about Con-Grab and "The Establishment", because people of "the left" don't "get it", not at all.
And I'd really hate to be repeatedly "911"d by a bunch of leftie-anti-suicide-bots. Or worse, SWAT'ted.
-
-
Wednesday 29th November 2017 20:03 GMT Anonymous Coward
Re: I wonder how many false positives they're getting...
Bob may not be the smoothest corespondent, but he's got a point. We Trump-voters are anything but depressed! Quite the opposite, we're all smugly satisfied at the results our votes have wrought thus far, with much more coming, oh yes. ;-/
It's the hard-core leftists who may bear suicide-watching. Most of the poor dears still haven't processed the new reality yet, tsk. For some it may never be possible, but there's still hope that those people might "learn to live with it," given enough care, and group therapy perhaps.
-
Wednesday 29th November 2017 21:33 GMT Pompous Git
Re: I wonder how many false positives they're getting...
"Bob may not be the smoothest corespondent, but he's got a point. We Trump-voters are anything but depressed!"
Why would they be? The economy is looking positive for the first time in a long while. Trump's tweets are frequently hilarious. What's not to like?Disclaimer: The Git and his son are both slightly left of centre politically. We are unabashed admirers of Trump and just as amused by the posturing of those further to the left than ourselves. Most amusing according to the Gitling are those who will no longer have anything to do with him because Trump!
-
-
-
-
-
Wednesday 29th November 2017 08:53 GMT rmason
Judging by..
Judging by the outpourings we see from commentards on here and my own experiences of them on facebook et al I don't think his supporters are disappointed.
Many seem to lack the intellectual capacity to see anything but what they want to see, just as many are overjoyed because he's doing exactly what they want and he says exactly what they think.
It's a bit terrifying. They haven't been "tricked" they genuinely are knuckle-draggers. Thick, backwards and bigoted.
-
Wednesday 29th November 2017 15:57 GMT bombastic bob
Re: Judging by..
"they genuinely are knuckle-draggers. Thick, backwards and bigoted."
As compared to many people on the LEFT, who are genuinely CONTROL FREAKS, SNOBS, and ARROGANT, that are COMPELLED to CONTROL OTHERS through governments and "howler monkey" Saul Alinsky style intimidation, because you ARROGANTLY *FEEL* that you are SUPERIOR to everyone else, who must COMPLY to your "feely" whims like a bunch of BORG (or else they're "knuckle-draggers").
Brainwashed, indoctrinated, FOOLS! Or, in other words, COMMUNISTS.
(there, I said it. happy?)
-
-
Thursday 30th November 2017 16:17 GMT Anonymous Coward
Re: Judging by..
Yeah Bob, go easy on those caps! You're addressing sensitive, caring souls who experience genuine pain when conservative troglodytes like you stomp in here wearing hob-nailed boots.
What you want to do is keep it cool, and delicately pick apart their carefully constructed world view, piece by misshapen piece. It really bugs them! Cognitive dissonance I guess.
That's how I achieve those near three-figure downvote totals. It's a dirty job, but someone's gotta do it! ;-/
-
-
-
-
Wednesday 29th November 2017 18:34 GMT Captain DaFt
Re: I hope the Facebook AI becomes sentient...
... and after seeing what it is, commits suicide.
Bonus points if it takes the entirety of Facebook down with it!
Then Twitter would go down too, since everyone on FB would jump there to complain.
World production/morale soars, accidents drop dramatically since no one's glued to their smartphone... It's win/win/win!
-
-
Wednesday 29th November 2017 09:31 GMT Anonymous Coward
It's not difficult to work out how the AI detects how you spend your life nurturing your ego by developing personality, making friends and life choices only to realise that your sole existence now is to garner attention and likes in a self serving environment where like minded people attempt to do the same in some sort of popularity content for which you see no escape because you think you will be alone in the world and lose all your friends.
Welcome to the social media prison, now what can we sell you today?
-
-
Wednesday 29th November 2017 09:58 GMT Anonymous Coward
Ridiculous trying to automate this
All it can do is basically what amounts to keyword search, but it can't really derive emotional state from them. How could it tell the difference between a serious suggestion of suicide and a sarcastic one? I know some people who have a broken sarcasm detector who can't tell the difference, and are forced to infer it from how others respond to a sarcastic statement.
The only way to address it is to have a way for people to flag statements/videos they are concerned over, and have humans at Facebook quickly respond. That's if you believe Facebook has a responsibility to address it at all - I'm not sure why it is their problem... Do the owners of tall buildings take responsibility for quickly detecting someone standing on a window ledge threatening to jump? No, they rely on passers by noticing and calling 911.
-
Wednesday 29th November 2017 11:08 GMT Anonymous Coward
Re: Ridiculous trying to automate this
Possibly not. Certain physical illnesses that effect or can often lead to self harm or suicide can pre-despise people to delusional episodes. How the software would be able to tell real posts from fanciful or imaginative lies is the difficult part.
However, educating friends and family to look out for each other should they receive somewhat strange communications would be beneficial.
Anon, because this actually did effect my family recently, and sadly the seemingly entirely fictional messages and stories were taken for real by the family, and the delusional episode entirely missed.
-
Wednesday 29th November 2017 10:00 GMT Lee D
To be honest, the last thing someone who is depressed enough to resemble someone suicidal wants is to be asked if they're suicidal.
"From what we can tell, it alerts human handlers to intervene if you're sounding particularly morose on the social network, prioritizing reports by friends that you're acting suicidal."
How does that work? "Hi, this is Facebook, we think you've been really depressed since you posted that your boyfriend slept with your sister, and your mate John told us all about you trying to get yourself drunk last week over it, would you like to talk to us about it?"
Given the amount of "Kill me now" kind of posts that aren't meant to be serious, the inherent (cough) reliability of AI, the potential for people to use this to PRESSURE people into committing suicide (by reporting their posts as being indicative of suicide), etc. this can only go badly, I feel.
You know what, I'm not even sure that Facebook should be doing anything but their job. Act on inappropriate posts on their social network.
Trying to second-guess deep psychological medical conditions with fatal consequences is probably beyond the scope of the EULA.
-
Wednesday 29th November 2017 10:03 GMT Anonymous Coward
Don't go there
I grew up in a major city with a population in the millions. I liked going to pubs, clubs and parties. In my experience the number of true friends that you make can be numbered on one hand.
The notion that you can have hundreds of "friends" is false.
If you want to mitigate suicidal thoughts then avoid social media. Talk to your true friends when you need help.
https://www.youtube.com/watch?v=KEXQkrllGbA
-
Thursday 30th November 2017 13:20 GMT MyffyW
Re: Don't go there
Some learned wonk on the Today programme the other morning suggested social media provides the same sort of dopamine hit as any other addictive behaviour. Having observed my own mood rise and fall based on likes and upvotes I'd tend to agree. So Facebook really is the last place a clinically depressed person should be. Still, if they were there anyway maybe a pop-up that said "Why don't you switch off your PC and do something more interesting instead?" would help...
-
-
-
Wednesday 29th November 2017 11:44 GMT Pen-y-gors
Re: Other applications?
You may have something there. Not so much sending round the re-education squads as at least identifying them for future action. Re-education would be a waste of time and money - if they haven't clocked by now that Brexit is a really, really bad idea, then no amount of 'education' is likely to work. (I liked today's comment from someone - instead of spending £50 billion on Brexit, let's dump Brexit and spend it on the NHS)
But if we have a list of the names then once Brexit has been canned we can decide what to do with them. There are many options:
1) Send them daily postcards and tweets saying 'You lost, get over it'
2) Ditto "It's the will of the people"
3) Banned for life from voting, due to clear and demonstrable mental incapacity
4) Transportation for life to New Botany Bay (East Falkland)
-
-
Wednesday 29th November 2017 10:18 GMT Brewster's Angle Grinder
Pre-GPDR, the Samaritan's Radar app, which attempted similar things on Twitter and was eventually pulled, got a lot of push back, some of it on data protection grounds. I can't remember the specifics, and I didn't see any argument I thought held up, but there was a non-trivial case against it.
So maybe their lawyers genuinely take a similar view. Maybe they think it's not worth the fight and the negative PR. And I'm sure Facebook would like loser data protection laws, so maybe, in part, they're using it as a wedge to weaken our laws: see the nice shiny things you can't have if you want privacy.