Assembler, the platform against false images
Detect digitally modified images It is essential today. And we are not referring to changes in color, exposure or saturation, but to those who completely change the reality that they had originally captured. For example, placing someone who was not there, modifying gestures or other worse things.
Therefore, almost any technology already works on mechanisms that help this detection of fake news Sometimes, the work is done manually with a team of people who verify the authenticity of the images. And on other occasions applications that analyze each pixel are used.
In that last group is Assembler, an experimental platform that Google Research and Jigsaw are jointly developing. This, through the use of different tools, is able to detect the most common effects that cause the use of different editing techniques: changes in colors, excessive noise, brightness, etc. And, to hide these changes, the most common thing is to “dirty” something the image to give credibility.
Well Assembler would be able to detect all this and even images or parts that could have been generated through artificial intelligence. Which would mean an important help for journalists or any data verifier.
Through a quick analysis, the screen would show the areas where the modified elements could be. Even so, although the system could be very useful to cover news that explodes in a matter of minutes, additional work will always be necessary as modified images are not always used in fake news.
A first anti fake news filter
The usefulness of Assembler and other similar solutions go beyond what allows different users on a personal level. These systems will also be very useful in the immediate future as the first filter for social networks. Automate and brake from the original upload process would help.
The only problem is to see to what extent other types of editions that are not looking for fake news could be affected. This is one of the main fears. Not so much in services or networks such as Twitter or even Facebook, where such solutions are really necessary, but in other Instagram types.
If you are interested in the topic, on the Jigsaw page in Medium they explain in a little more detail how they apply all these techniques to their experimental platform, the universities involved in the development and some extra details that give a clear perspective on the topic of disinformation and how it goes beyond simple fake news