Juan Villoro
Technology determines our lives, but only a few know its true scope. Convinced that all change implies improvement, digital business promoters and fans of the future sleep soundly. The rest of us have something to worry about.
Every time I talk about the subject with a specialist, I feel like I’m entering the cockpit of an airplane: I want the pilot to minimize turbulence, but I fear that his diagnosis will be terrible. Knowing a lot does little to tranquilize me.
I experienced this in a conversation with Carlos Coello Coello, an expert in computer science, researcher at Cinvestav, and member of the National College, who rigorously studies a phenomenon that almost all of us approach intuitively.
The digital revolution arrived without a pedagogy to understand it. Coello Coello belongs to the select minority who know what is happening. After listening to him, I admired the state of mind with which he approaches a field that is both stimulating and full of threats.
The term “artificial intelligence” was coined in 1956 and has been used for years to describe computer programs that manipulate symbols, simulating human reasoning. Until the 1990s, specialists had a detailed understanding of the steps the machine followed to produce results. Everything changed with generative artificial intelligence, which solves tasks for which it is not specifically programmed. The machine does not start from scratch, and to that extent, lacks invention. Still, it uses data with “unprecedented artificial creativity,” according to Leonardo Banh and Gero Strobel, researchers at the University of Duisburg, Germany. Development has become so rapid that programmers obtain results without specifically unraveling the generative sequence that makes them possible.
We are facing the partial autonomy of AI. This function still has errors, known as “hallucinations,” which Banh and Strobel define as “results that seem plausible but are irrational” (Electronic Markets, 2023).
Coello Coello raises an essential question: in whose hands is all this? Although there are millions of developers, the design of LLM (large language models) code depends on a handful of experts. According to David Luan, who led the team that created ChatGPT 2, there are no more than 150 people. Coello Coello raises the figure to 200 and adds that, in total, the experts capable of “building and conceptualizing the training of a frontier model” may number as many as 500.
It is not known how many there are in China. Rumor has it that the four most important ones had their passports taken away. The country that was once ruled by the Gang of Four now depends on a quartet of scientists treated as hostages.
LLM code creators have sparked signings that rival those of the NFL. A rookie like Matt Deitke, born in 2001, the year of “2001: A Space Odyssey,” didn’t have to finish college to be hired for $250 million.
Veterans earn even more. The Wall Street Journal reported that Andrew Tulloch, founder of Thinking Machines Lab, received an offer from Mark Zuckerberg of more than $1 billion in salary and stock, payable over a six-year period. Previously, the owner of Meta had hired Shengjia Zhao, a developer at OpenAI, with an advance of nearly $100 million.
On September 4, Donald Trump dined with tech moguls (minus the self-exiled Elon Musk). Before dessert arrived, Mark Zuckerberg announced that he would invest $60 billion in AI. Part of that will go toward signings that make Real Madrid’s “galácticos” look like a grocery store.
The dance of the billions heralds a progressively divided world in which the owners of technology necessary for communal life (i.e., “digital”) will manipulate consumers.
Most worrying is that knowledge will be concentrated to the maximum: the advancement and control of generative artificial intelligence will be in the hands of the “Top 200.”
The situation is reminiscent of ancient China, where the mandarin caste decided how the entire society would function. The elite understood that knowledge is a resource of domination. However, it also showed that you can be both erudite and crazy.
The fate of the planet depends on the mental health of 200 people.
Further Reading: