Facebook trains AI to help prevent suicides

San Francisco - Facebook says it is stepping up the use of artificial intelligence (AI) to identify users of the leading social network who may be thinking of suicide.

Software will look for clues in posts or even in videos being streamed on Facebook Live, then fire off reports to human reviewers and speed up alerts to responders trained to help, according to the social network.

"This approach uses pattern recognition technology to help identify posts and live streams as likely to be expressing thoughts of suicide," Facebook vice president of product management Guy Rosen said in a blog post.

Signs being watched for are said to include texts by people or comments to them, such as someone asking if they are OK. 

Facebook already has tools in place for people to report concerns about friends who may be considering self-harm, but the software can speed up the process and even detect signs people may overlook.

"There have been terribly tragic events - like suicides, some live-streamed - that perhaps could have been prevented if someone had realised what was happening and reported them sooner," Facebook chief executive Mark Zuckerberg said early this year in a post focused on building global community.

"Artificial intelligence can help provide a better approach."

Facebook is rolling out the artificial intelligence tool outside the US and plans to eventually make it available everywhere except in EU countries, where data usage is restricted by privacy regulations.

Facebook has been collaborating with mental health organisations for about a decade on ways to spot signs users may be suicidal and get them help.

* Sign up to Fin24's top news in your inbox: SUBSCRIBE TO FIN24 NEWSLETTER

ZAR/USD
16.98
(-0.23)
ZAR/GBP
21.23
(-0.06)
ZAR/EUR
19.15
(-0.17)
ZAR/AUD
11.80
(-0.12)
ZAR/JPY
0.16
(-0.25)
Gold
1774.31
(+0.03)
Silver
18.02
(-0.04)
Platinum
808.00
(+0.25)
Brent Crude
42.78
(-0.79)
Palladium
1914.01
(+0.62)
All Share
54521.90
(-0.17)
Top 40
50179.89
(-0.26)
Financial 15
10150.02
(-0.64)
Industrial 25
76554.73
(+0.52)
Resource 10
50138.02
(-1.24)
All JSE data delayed by at least 15 minutes morningstar logo
Company Snapshot
Voting Booth
Please select an option Oops! Something went wrong, please try again later.
Results
I'm not really directly affected
18% - 1609 votes
I am taking a hit, but should be able to recover in the next year
23% - 2115 votes
My finances have been devastated
35% - 3173 votes
It's still too early to know what the full effect will be
25% - 2295 votes
Vote