We at APPLY teach computers to see and understand our world with the help of artificial intelligence algorithms and neural networks. Teaching computers helped us to understand how we, humans, interpret and process information. It also helped us to understand what kind of information we have around us. It is our strong belief that this knowledge is vital for our ability to identify media threats and to develop measures of fighting them.
There is no doubt that not only our physical environment but also information around us shapes us as personalities with the course of time. It starts in childhood. Imagine a first-grader, let us call him Alex. His mother thinks that Alex is too young, so he is not yet ready for any challenge in his life. His father, though, believes that Alex must be given more independence so that he can find out on his own what consequences his actions may bring about. There can be no unanimous opinion as to who is right in this situation, but one thing is clear: two parties are struggling to influence the third party. Their goal is to make Alex adopt a specific opinion and act accordingly. What is it if not propaganda?
Let us imagine Alex choosing his mother’s cautious approach. It will create in Alex certain proclivity for dealing with challenges, as well as life in general, in one way and not the other. Humans are predisposed, prejudiced or even biased in consuming information and passing it forward. No matter whether all this comes from upbringing, environment, or other experiences, our brain works like this. Confirmation bias, gambler's fallacy, post-purchase rationalization, and even bandwagon bias make us who we are.
Through the years, we undergo a lot of influence, absorbing and accumulating the effects of it. Let us fast-forward some forty years to see Alex as a grown-up and a news producer at a respectable TV channel. He is the one who decides which topics will be covered, how much time will be allotted to a certain matter, which people will be hired, and how new journalists will be mentored – thus shaping the source from which all the viewers of his channel will receive information. Consciously or not, he distorts information, and his point of view influences others.
Let us make things more complicated. Say, the biased Alex is an open source intelligence analyst, whose job is to monitor the political environment in one of the neighboring foreign countries. He prepares reports for the people who will use this information as grounds not only for foreign policies but also for action planning. It can happen that Alex’s confirmation bias or observational selection bias will play a role in how these reports will be worded, what facts will be included in them, and, consequently, those biases will influence the mentioned policies and actions in the long run. What if Alex will consciously push for the certain tonality of those reports because of his own beliefs?
It is common for humans to give under information pressure, meaning that the more often something is repeated to a person, the more likely the person is to accept it as truth, regardless of any logic or common sense. Moreover, one does not need to live in North Korea to be under such an information attack: it is anywhere on the planet that a person can be exposed to a purposely-crafted information stream from media driven by an unknown agenda. There are many aspects making a person vulnerable to propaganda:
not speaking the local language;
speaking the language, but feeling alienated from the family or his or her ethnic culture;
lack of education;
being a member of a certain community.
Thus, we should focus on avoiding subjective input as fully as possible, getting more effective and efficient, and all that – despite the obvious inability of the human mind to keep up with the volumes of content in mass and social media.
Monitoring social and online media has never been easier, even with each person being considered as a subjective medium. Another notable fact is that only around 45 % of the world’s population have access to the internet, which is less than a half. About two-thirds of that 45 % live in countries that have internet censorship and surveillance programs in place. Those people most probably consume TV and radio broadcasts, or any other visual content streaming. And those are notoriously difficult sources of data to process.
Picture this: Alex and his team are tasked to monitor TV broadcasts, patterns, and pattern shifts of what is broadcasted, the sentiment of broadcast, tonality of talk shows, or jokes in humor shows, etc., because all this gives insight in the respective country’s political direction, agenda, and priorities. Thus, Alex and his team watch different channels 24/7. They make notes, take screenshots, count the appearances of certain public figures, transcribe announcements, etc. Can the team, being so busy recording and observing, notice a shift of attitude towards one of the neighbors? Probably yes, if that shift was made clear shortly before. Yet, to detect more subtle attempts to manipulate the public, i.e., attempts undertaken with a view to the future, a retrospective analysis would be needed. We can assign another team to perform this analysis, that is, to watch the relevant talk shows and news again, provided that those have been recorded or have been in the focus of Alex’s team. However, even if they have, they need to be processed and structured in a certain way, otherwise, the efficiency of the analysis will be compromised, as the analysis will take an enormous amount of time.
Above all, TV is incredibly influential when it is the main source of information. This raises another notable concern. Alex could have something like “shelf life of an analyst” problem. Regardless of how well Alex is trained, while working within the same paradigm for a long time, he will be influenced by the very same propaganda he is analyzing, which will drop the efficiency of his job and the quality of his output.
Imagine there is Mark. Mark is able to analyze huge amounts of information. Mark does not forget anything and is capable of following through without a slightest status quo bias or negativity bias. Nor does Mark lose his focus or have any political or religious affiliation.
He accounts for spoken and written texts in the news and TV shows, understanding more than six languages, identifies objects and scenes featured in promoting a certain topic, points out known and unknown persons, provides demographic information about them and characterizes their attitudes, describes audio and visual effects used to convey the message. All of this is combined and analyzed semantically, resulting in a complete report, which contains data on:
who said what and how;
where the information originated or where it was first mentioned.
Mark’s reports are not static, they are updated as soon as new broadcasts about the relevant topic appear.
Mark is a computer. To be fair, he is not just a computer, for he could also be described as a media data mining system, which has certain useful artificial intelligence abilities.
“Big data” is a buzzword nowadays, and there is a reason for that. It can tell a story about us, about how we live, what we like, what we hate, what is our shopping behavior, etc., as all that is easily extractable from social media, or bank statements, or loyalty cards.
Yet some sources, like TV, cannot be analyzed without advanced solutions, which we did not have until now. Whereas now, in addition to content analysis, we can even understand who are the consumers of information coming from such sources.