Surface Ripples

AI – A Downside

A news article(that appeared in July, I think)  reminded me of something I posted on artificial intelligence (AI) earlier this year, and whether it was a good or bad thing. 

In the article, social media companies were using AI to remove postings that might contain offensive, harmful or indecent material.  The problem was, the AI couldn’t distinguish between content that was harmful, offensive or indecent and content which showed war crimes or acts that violated human rights being committed.

This is one of the cases we find of “the law of unintended consequences.”  Social media companies have been under the spotlight for allowing posts that contain hate speech or other content likely to be considered offensive.  The fact that a lot seem to get through still is neither here nor there.

In response to the pressure to moderate effectively content posted, social media has naturally turned to one of the newest tools in the proverbial toolbox.  Given the millions of postings every day on the likes on Facebook, “X”, Instagram, Snapchat, and goodness knows how many others, we have to accept that no human being or group of human beings (unless numbering in the millions) can successfully moderate content, removing that which is gratuitously and unnecessarily offensive or harmful whilst still leaving in place content which can be used in evidence in war crimes or human rights abuse trials.

One answer is simple.  If someone is a witness to an event that might constitute a war, crime or abuse of human rights and photographs or films it, they need to have a site to which it can be sent. Rather than the “standard” social media sites, they may need to store it on their own personal cloud, Dropbox or whatever personal storage, they may have already.

The results of one particular war crimes content posting being removed was that a refugee applying for political asylum and giving danger to his own life as a reason for doing so lost his proof as, unfortunately, the video footage that showed him in the place where the incident took place had been erased by AI.

We are only just beginning to come to terms with AI’s strengths and weaknesses. Like many other aspects of life which we now take for granted (e.g. cars, trains, aircraft, broadband internet and others) it will be a case of time showing us what needs to be done.  However hard people may try, they cannot legislate for everything all at once.

At best, they can look at the lessons of the past and decide where to apply them in the case of AI. 

I’ve spent more than half my life delivering change in different world markets from the most developed to “emerging” economies. With a wealth of international experience in international financial services around the world running different operations and lending businesses, I started my own Consultancy to provide solutions for improving performance, productivity and risk management.  I work with individuals, small businesses, charities, quoted companies and academic institutions across the world. An international speaker, trainer, author and fund-raiser, I can be contacted by email. My website  provides a full picture of my portfolio of services.  

Labels: Crisis Management, Customer Care, Productivity, Risk, Strategy

posted by William Martin @ 14:46