AI, the New Myth
December 4, 2017, 4 min to read
AI is fascinating. AI can do a lot of things, but AI cannot do everything. And AI is certainly not everything. To feel compelled to utter such platitudes sheds light on the strange tone that the debates about the consequences of AI often take.
Massive extinction of employment, totalitarian dystopias, merciless killer robots on the one hand; immortality and other small perks in (over)abundance on the other… AI, in these scenarios, sounds like a giant myth. Even if it looks towards the future rather than time immemorial, AI still shares the themes, structure and goal of a myth.
The themes: broad existential questions that have always been present. You can find immortality in the ambrosia of Olympus just as in the Singularity imagined by Ray Kurzweil, and the fear of unleashing evil forces in Pandora’s box just as in Elon Musk’s televised anxieties.
It is very convenient to associate AI with endless themes for at least two reasons. First because AI is so complex that it is incredibly easy to say that everything, hopes and fears alike, depend on it. Then because this field, through the figure of the robot, falls in line with a long series of stories - the living statues from antiquity, the Golem of the Jewish tradition, Frankenstein’s creature or Asimov’s androids - that question the essence of humanity through the relations we maintain with our creations.
The structure: what matters in a myth is the internal coherence of the narrative, not the rational underpinnings that could make it possible. To me, this is one of the characteristics of a famous essay penned by Nick Bostrom, Superintelligence, which offers a meticulous analysis on the prevention of risks that an intelligence far superior to ours could pose to humanity. A brilliant, erudite book… whose fundamental flaw is to expedite the study of the concrete steps that could enable - or not - such an outcome. Superintelligence of the future and magical forces of Olympus are of the same kind.
The goal: articulating broad morals for the present. Beware of hubris, are warning us both Icarus and alarming forecasts about the damages of AI.
This is where it gets extremely interesting: what are the contemporary debates around AI questioning us about? In my view, this technological revolution serves as a welcome mirror to our society. And if there is anything to be mad about, it is that today’s roots of the anticipated problems that are already attributed to AI have never been tackled before.
Killer robots? Let’s talk right now about the regulation and control of drones.
Massive layoffs? For years we have been passively observing inequalities due to unemployment and unfilled vacancies, and we regularly call for the advent of a system that would train everyone, and not just the most qualified, on a lifelong basis.
Biased algorithms that favor white / rich / qualified / men / hailing from developed countries…? They just reflect, and reproduce, inequalities that have been so often pointed at - but less often fought.
A surveillance society? We did not have to wait for powerful machine learning algorithms to implement large-scale spying programs.
At FABERNOVEL, we will never forget to consider potential harm caused by AI, to take ethics and values as principles of action, not as theoretical gibberish. But we will not forget either that the real threat that AI poses to society is to amplify existing problems, for which it is seldom the only solution - and sometimes, not one at all.
We have always had great myths - the mistake would be to believe that in the 21st century, we do not need them anymore, that progress and scientific rationality have made them vanish. Nevertheless, hidden as they are among essays from futurologists rather than epic poems told orally by one generation to the next, we have forgotten their nature and function: by pushing the limits of what is imaginable, they invite us to take a step back and reflect on present times - but are not to be understood at face value. We should not mistake goads for goals.