deep fake

We are used to the idea of manipulating images to insert something that wasn’t in the original, and children at school are now instructed, it seems, in techniques for spotting such a manipulation, as, for example, comparing the shadow of the suspect person or thing  with the shadows already in the picture. If there is an insertion then the shadows won’t match.

 The deep fake is even more insidious because computer technology now provides techniques for seamlessly inserting contrived audio or video into real audio or video, the final result being apparently completely authentic.

 Given enough material to work from, a computer program can engage in deep learning and analyse sound or movement into small building blocks which can be rearranged to create artificial chunks. An individual can be made to say anything you want them to say or do anything you want them to do. This is disconcerting to say the least, because it makes it impossible to distinguish fake news from real news. It has been traditional in news publishing for news organisations to say We will give you the news and it will be accurate. The worst thing a journalist could do was to introduce an error of fact into a story.  Now we need such publishers online to say We will give you the news and it will be not be fake!  And keep their promise. More than ever we need to consider the trustworthiness of the source of our information.