The news that Facebook and its sister platforms were being forced to do something about ‘revenge porn’ highlighted how early in the life of Artificial Intelligence (AI) we actually are.
Mr Zuckerberg wrote in one of his famous blogs that ’we are focused on building a community that keeps people safe. That means building technology and AI tools to prevent harm’.
Which is fine, but technology itself cannot spot ‘revenge porn’ just as it cannot spot bullying, the language and skill is too subtle (‘your mother doesn’t visit you because she doesn’t like you anymore’). In fact, the person in charge of tackling revenge porn at Facebook rather disagrees with Zuckerberg anyway.
What in fact happens is that humans, who are also on Facebook, are asked to hit the ‘report’ button if they see something dodgy. These are then screened by humans who also work for Facebook. The post and probably the user is then blocked.
Then comes the clever (ish) bit. They use photo matching technology to make sure that if someone tries to share the picture, it is blocked.
So, not exactly AI.
More your reasonably common or garden look-up technology.
The scepticism around AI, and a growing list of other over-hyped technology, is now growing almost as fast as the hype itself. And while Facebook is probably furthest behind in AI, according to Richard Windsor, for the moment AI is more artificial than intelligent.
Still, the rise of AI is underway, whether in its purest form or as a way of making sure that automation is effective.
Indeed, it now seems that having a human involved is something of a luxury.
Governments are looking at ways of minimising the impact of AI on jobs, with a ‘made by humans’ tag on goods. And luxury brands are promoting premium services as being supported by humans in their call centres.
There are, of course, areas where AI, in its support of intelligent automation, makes sense. Google invested huge amounts of money in AI or ‘deep neural networks’ in order to make their processors more effective and save them many millions in not building even more data centres. They did this because voice controlled devices are producing unparalleled quantities of data.
For the moment, the most that AI can boast is very, very fast look-up and processing. It will be a while yet before we see proper applications.