Faster than suicide – Stopping suicides with artificial intelligence?

10.000 people take their own lives in Germany every year

Worldwide there are even more than 800,000 people a year. The World Health Organization (WHO) mentions this figure in its “World Suicide Report”.

Every incredible 40 seconds somewhere in the world a desperate man ends his life

Not included in the statistics are the suicide attempts. It is estimated that this number is in the millions. Men commit suicide significantly more frequently than women across all age groups. The statistical average age of suicides in 2015 was 58.3 for women and 57.1 for men.

Sad Selfie – Trend

Last year, 33-year-old musician Jared McLemore doused himself with gasoline and set himself on fire. A 12-year-old American woman hanged herself in the garden. As different as their ages and motivations may have been, they had one thing in common: both transmitted their deaths live via Facebook.

I suspect that dying is also being digitalised and that suicide selfishness is entering as a “modern variant” of the farewell letter.

Death is great.
We are His


wethinkin the middle
he dares to weep
(Rainer Maria Rilke)

The causes of suicides are manifold

Statistics can only answer – where, when and by whom a suicide was committed. No statistics can provide the answer to the “why”. This is why relatives often remain disturbed and questioning. If a video of the dying person is then circulating on the Internet, one can hardly imagine the pain of those left behind. This makes it all the more understandable that ways and means are being sought to prevent suicides where they are announced and even streamed live – in the social media.

Being faster than suicide with artificial intelligence

Chatting, posting, uploading videos and licking – Facebook has the largest mountain of data there is about people – 2.1 billion users belong to the network. It is obvious to look for the “why” of suicides in this mountain of personal data. Behind the statistics of suicides are hidden personal fates and life stories that are often communicated beforehand in the social media.

Facebook uses artificial intelligence to identify posts that suggest that the author is suicidal.

An algorithm that finds typical patterns

As a data basis for the evaluation, all contents – such as posts, comments and videos of the users – are to be scanned for alarming keywords. As examples, Facebook cites “Are you okay?” or “Can I help you?” in its press release

As soon as the pattern recognition system sounds the alarm, the mail is forwarded to a team of experts who – depending on the test – contact the persons concerned. The team also informs the authorities, psychological emergency services or relatives as appropriate.

As an advocate of the suicide algorithm, one can argue with a moral obligation. Preventive interventions as a duty, which arises from the possibility of interpreting data to save lives. To exaggerate – to have the opportunity but not to use it could be interpreted as failure to provide assistance.

Privacy activists are alarmed

Critics argue that, on the one hand, not only conversations of potential suicides are scanned and evaluated, but those of all users. Secondly, the way the analysis works is not sufficiently transparent.

It warns of the consequences of mistakes: when the artificial intelligence errs and falsely accuses a person of intending to commit suicide.

Imagine that somewhere in the world someone unsuspectingly opens a front door and finds a police detachment and psychological service in front of it. Stigmatisation for those involved and affected could follow.

Priorities and transparency

The moral question behind the discourse is, in my opinion:

What has higher value? The possibility of saving lives – or the right to privacy?

If the majority allow insight into their personal conversations and posts for the protection of a minority – does not this give them the right to transparency? About how and how often it is analysed? Is a company like Facebook allowed to operate here in secret?

Kommentar verfassen