Google put its version of artificial intelligence, which it calls Gemini, online in our area last week. I assume he discarded his paronym “Genesys”. In one of the latest films Terminator, this is the name of the software programmed to take control of the planet. In half a century, perhaps sooner, with things moving so quickly, someone will write the genesis of the new era. The aim will be to determine who were the demiurges of artificial intelligence (AI). How was the decision to let her out of the laboratory made? And when people aware of the danger wanted to avoid the worst, why did they fail?
The year 2023 will be retained as year zero. The name of Sam Schillace, a Microsoft executive who financed the first born of the AI age, ChatGPT, will surely be entitled to a quotation. This, reported by the New York Times, written when he was worried about being beaten by Google: “Speed of execution is more important than ever,” he proclaimed. It would be an absolutely fatal mistake, right now, to worry about things that can be fixed later. »
These things to be repaired later being the propensity of the AI, in its responses, to leave the factual terrain to begin to invent, to confabulate, plausible assertions. Its propensity, too, to spontaneously leave the framework that had been set for it, without its designers understanding exactly why. Which pushed cutting-edge researchers, including our national Yoshua Bengio, to admit that by creating the conditions for the birth of AI, they had generated Frankenstein.
One such skeptical scientist is Stuart Russell, a computer scientist at the University of California: “It’s almost like you’re deliberately inviting aliens to land on your planet and having no idea what they’re going to do when they arrive here, except they are going to take over the world. »
Last November, the board of directors of OpenAI, an NPO overseeing ChatGPT, wanted to apply the brakes. Under the leadership of its chief scientist and co-founder Ilya Sutskever, the council fired the project’s kingpin, Sam Altman, considering that he could no longer be trusted to guide the evolution of AI. for the good of all humanity.” But since Microsoft was OpenAI’s main funder and “the good of all humanity” is not the driving force of its economic activity, the giant immediately hired Altman and invited the 770 employees responsible for ChatGPT to join it. No less than 700 said yes.
The board changed its mind, rehired Altman. The scientist Sutskever said he “deeply regrets” having wanted to fire Altman. The attempt to master AI has failed spectacularly, under the crushing weight of the search for profit. “ We are so back », proclaimed one of the leaders of the pro-Altman mutiny, savoring his triumph. Did he know that he was paraphrasing the famous “ I’ll be back ” of Terminator ?
But perhaps we cannot explain the headlong rush towards AI only by the dynamics of competition between companies (and States) to be the first to arrive, and therefore the first to be served, in money and power.
Journalist Ezra Klein interviewed for the New York Times several of the key architects of this revolution. “I ask them the same question: “If you think calamity is this possible, why continue to work on it?” Different people have different answers, but when I press, I discover answers that seem to reflect the point of view of the AI itself. Many — not all, but enough that I can confidently say so — believe that they have the responsibility to usher in this new form of intelligence into the world. »
In a gripping essay, author Paul Kingsnorth notes that at the center of every culture there is a throne that requires a god. However, the West having dethroned its ancient deities, the throne remains to be taken. He emphasizes that thinkers of the industrial revolution already predicted that the mechanical and individualistic world would generate, between Christ and the devil, a spiritual being adapted to our era. One of them, Rudolf Steiner, called him “Ahriman”. Kingsnorth recalls that Marshall McLuhan already declared that the emerging computer world constituted the “central nervous system” of what he believed would become a new consciousness.
The term “Ahriman” has been used for a quarter of a century by transhumanists, who believe that we will soon witness a fusion of man and machine. This process has just taken a leap forward with the implant that one of Elon Musk’s companies has begun testing on human brains. There would therefore be, if we follow this logic, a convergence between pure scientific research on AI and the invocation of a power lurking in its circuits and its data.
Kingsnorth summarizes: “Something is crawling towards the throne. The impulse that now shapes and reshapes everything, the earthquake born through the wires and turns of the web, through the electric current, the touch screens and the helmets: these are the pains of contractions. The Internet is his nervous system. His body takes shape in cobalt and silicon and in the great glass towers of cities. His spirit feeds on the constant 24-hour flow of your minds and mine and the minds of your children and your countrymen. No one is asked to consent. No one even has to know. It happens anyway. The great spirit is being built. »
“Does God exist? » asks the scientist, big man at Google and transhumanist author Ray Kurzweil. He replies: “Not yet. »