Omnipresent and omniscient, big data looks like the big brother of 1984 by George Orwell. We generate floods of data through our Internet searches, our online interactions, our geolocation data… And once sorted and analyzed, algorithms in the form of supercomputers get to know us better than ourselves.
The spectacular computing capacities of artificial intelligence have developed its powers of prediction and suggest a new way of approaching the collective future. There is a great temptation to let him make decisions for us … In the light of the presidential campaign, let’s come back to this digital entity with infernal powers that should – perhaps – be put back in its place as a simple machine. No longer call it big data but “datagod”. The proof.
Omnipresent, omniscient… and omnipotent?
He is everywhere and he knows everything about us… Big data is the flow of all the data that we leave online, whether it is health data, the data that Facebook collects from its users, the whole of our interactions, photos, emails, emojis, purchases… He knows absolutely everything about us but we can’t see him. “Like a god, he can rule our lives or help us in our lives but he is strictly invisible, observes Sylvain Louradour, author of the book. The New Now of
the Netexplo digital innovation observatory. The closest physical thing to big data are servers ”.
By gleaning these millions of digital data, he ends up knowing more about us than ourselves. “If we crossed the data of our purchases with our medical consultations, our trips, our relationships …, we would see things about ourselves that we were probably not aware of, but which exist”, continues the analyst. And when we see that artificial intelligence is already capable of managing military or medical operations, we are not far from omnipotence. God, are you on my cell phone? Yes…
Faith in AI
It takes faith to rely on a higher invisible power. As we have said, artificial intelligence has computing and data processing capacities infinitely (and the word is weak) superior to that of humans. But it lacks all the other ingredients of human intelligence: emotion, critical thinking, the capacity for abstraction, common sense … everything that allows you, dear readers, to differentiate yourself from Siri, Alexa and OK Google.
“When we have to make a decision, don’t we tend to trust a portable god at hand, an entity capable of handling billions of parameters, asks Sylvain Louradour. But the decision we will make at the end does not need the objectivity of the data, it needs our own personality ”. This is where faith is needed. Especially since a large part of the data collected by Facebook, for example, is unusable, admitted Nav Kesher, data science manager of the social network in 2018 at the AI Summit in San Francisco. “Any attempt to predict the future via AI therefore seems doomed to failure, unless you settle for a partial future. However, our belief persists, ”concludes the Netexplo study. And, despite that, we continue to entrust a part of our existence to it.
A message decrypted by technoprophetists
Humans have created this digital god who now escapes them. “Big data is gigantic flows that are completely uncontrollable and incomprehensible”, describes Sylvain Louradour. The principle of deep learning is to program the machine to learn from its mistakes and to progress. She ends up deviating from what she was programmed to do and she arrives at results of her own. Because artificial intelligence does not work like our brains. “Like Pinocchio, the system at one point comes to life and escapes us”, points out the author of The New now.
In a religion, priests, rabbis are responsible for deciphering the meaning of the divine message, this is called hermeneutics. Likewise, the most advanced specialists in big data and AI “appear as oracles because they understand a little better than others what is happening in the machine,” notes Sylvain Louradour. But by interpreting, they orient the message according to their subjectivity. And we can see how an error in the interpretation of data can pose a problem in an electoral campaign.