An AI to distinguish between conspiracy theories and real conspiracies

AI is present all around us, invisible and ready to provide us with many services. It is this which helps us to find the shortest route to reach our destination, to take the best photo with our smartphone or even to find the most relevant results in our research.

There is thus a multitude of artificial intelligences and one of them has set itself the goal of helping us to distinguish between conspiracy theories and real conspiracies.

An AI to distinguish between conspiracy theories and real conspiracies
Pixabay credits

If you've had the opportunity to hang out on Twitter, Facebook, 4chan, or even Reddit recently, then you must have realized that conspiracy theories are rife.

Conspiracy theories, still very much in fashion

Covid-19, for example, has given life to a multitude of fanciful theories, theories going as far as asserting that the vaccine was intended to inject us with nanoparticles developed by Microsoft or that the virus had been patented by the Pastor Institute.

Most people of course can tell right from wrong, but not everyone. Even less at a time when certain political figures are blithely surfing the trend in order to rally as many people as possible to their cause.

This is a real problem, especially since the conspiracies have accompanied us throughout our history. In 1587, Queen Mary Stuart was accused of conspiring against Elisabeth 1st, which resulted in her being executed by three blows of an ax. A little later, in 1606, Guy Fakes was found guilty of the execution of the Powder Conspiracy. Louis XVI, despite his status as king, did not escape beheading either after being found guilty of “conspiracy against public liberty and the general security of the State” by the National Convention.

More recently, it was Ben Barka who was sentenced on board in absentia with Mohamed Basri, Omar Benjelloun and Mohamed Bensaid Ait Idder for having plotted against their king and attempted to assassinate him.

To read also: Deepmind has just unveiled a somewhat particular AI

How do you tell the difference between conspiracy theories and real conspiracies?

But then, in this context, how do you distinguish between a conspiracy, in the strict sense of the term, and conspiracy theories?

Timothy R. Tangherlini, a professor of literature and Danish culture working for the University of California, Berkeley, and Vwani Roychowdhury have just tried to develop an automated approach to enable us to distinguish the wheat from the chaff, in s 'relying on artificial intelligence.

To do so, they chose to focus on the narrative approach that generally accompanies conspiracy theories.

Unlike conspiracies, conspiracy theories most often result from collective storytelling. A group of people starts a rumor, it is then enriched by other individuals, until it forms a whole which seems, at first glance, perfectly coherent. It is therefore born from a collective writing work.

Read also: When AI takes an interest in Alzheimer's disease

A story of stories

The AI developed by the researchers is precisely able to identify narratives based on sets of people while determining their relationships. It thus helps to acquire an overview of a given problem and also makes it possible to distinguish between truth and fiction.

And so between the conspiracy and the conspiracy theory.

The AI has of course been tested with several popular conspiracy theories and in particular with that of Pizzagate.

Launched during the 2016 presidential elections, this theory indicated in substance that some relatives of Hilary Clinton had formed a pedophile ring, a network which had a pizzeria as its theater, all against a background of satanist rites to put a layer back.

Also read: What if AI helped us choose the best fish for our sushi?

A tool that could be of great help to us

The AI developed by the two experts made it possible to identify several strata at the level of the narrative and it thus highlighted the fact that the theory was in reality an accumulation of several pieces of information having nothing to do with each other. . By extension, she also revealed that removing just one layer of the narrative was enough for the whole story to fall apart. Proof that it was not based on anything concrete in reality.

The tool therefore seems efficient, and it could prove to be of great help in the future to fight against conspiracy theories, but also against disinformation campaigns.