AI: Utopia or Dystopia, Macho?

In April of this year, TED held its annual symposium to discuss advances in technology and design (mostly from Silicon Valley). The opening keynote address was given by Carole Cadwalladr, an investigative journalist who was subsequently sued. a talk in 2019. On that occasion, he spoke about the Brexit vote and the influence of Facebook – the network that most spread biased or outright false information. Cadwalladr had already written a report on the scandal of Cambridge Analytica in The Observer, but the TED conference was what billionaire Arron Banks used to sue her. The journalist's case has been a paradigmatic case of SLAPP (strategic litigation against public participation), a technique – which we Catalans are well aware of – in which a lawsuit is used as a deterrent exemplary punishment. Despite the harshness of the warning, Cadwalladr returned to the TED symposium to address the CEOs of Silicon Valley: this, gentlemen, seems a digital coup d'état.

Cadwalladr wanted to address, more specifically, Sam Altman (CEO of Open AI), who was in attendance. Because, according to her, AI was built on theft. To build AI, to create their product, companies ignored intellectual property laws. They steal our texts, our songs, our images, our voices. Absolutely all the reflection, art, "content," and interactions they've had access to because they were in the cloud. And they steal it with the goal of being able to recreate all of this synthetically, eliminating most of us. But for Carole Cadwalladr, the absurdity doesn't end there. Because if information is power, privacy is power too, Cadwalladr argues. She says this with full knowledge of the facts, given that during the years of the trial, her every digital move (and therefore her thoughts and intimacies) has been subjected to exhaustive scrutiny. However, privacy continues to lose ground: if we have virtually nothing left that belongs to the private sphere, using AI as a helper-confidant will eliminate the last bastions of privacy we still retained.

The journalist's inaugural conference also epitomized the radical nature of digital patriarchy. A woman on the margins of power, trampled by power, tried to convince those who monopolize her that they were being too reckless and aggressive with their companies. However, Carole Cadwalladr is just one of many women in the tech world with this spirit of resistance. As part of her journalistic project How to survive broligarchy, this summer Cadwalladr interviewed Karen Hao, which analyzes the situation in very similar terms to the book Empire of AI. Hao is another of the most relevant figures in specialized journalism in the technological world and is also highly critical of the current trend. Hao's book explains the evolution of the company led by Altman, which went from being a non-profit research organization (co-founded with Elon Musk) to becoming the company with the most dazzling projection on the planet. Open AI is now part of Stargate, a project with an initial investment of $100 billion (which could increase fivefold over the next four years). At record speed, Altman's company, backed by Japanese banker Masayoshi Son, has begun construction of the world's largest data center in Texas.

Hao rates the speech of the tech bros almost religious: the development of AI is like the search for the Holy Grail. According to Hao, AI should be trained only with a limited amount of data and for specific purposes (such as solving ecological or medical problems, for example). Instead, what these entrepreneurs have on the horizon is achieving an AGI (artificial general intelligence) that will entail an ecological and social cost that seems unaffordable. The disruptions we will see in the coming years will be unprecedented, because the scale of this (digital) infrastructure is also unprecedented. With Trump in government, everything is easy for companies pursuing AGI, regardless of whether they need a huge amount of drinking water (to cool the heat caused by so much computing) and electrical energy. If it seemed that we were going to leave nuclear energy behind, now the plans are different: in the United States they are recovering old nuclear plants and have accelerated uranium extractionAI needs all the energy we can produce and more. And if surviving AI will be complicated ecologically, it will be even more so socially. The disruption will go far beyond a painful disruption in the labor market, because what AI will call into question is the value of our ideas, our creations (but I'll leave that for another article).

Cadwalladr and Hao are radical in their precepts: we should ignore language models while they are led in this way, use free software whenever possible, and use messaging networks like Signal (which is not powered by AI and is led by another activist, Meredith Whittaker). These measures are difficult to implement, but both agree on one path that is unavoidable: the leading role of public opinion in demanding new laws and a slowdown in the pace of development. progress (which has never been so hasty). For now, there is a journalism giant that has joined the resistance. In 2023, The New York Times sued the OpenAI-Microsoft duo to train ChatGPT with your articles ignoring copyright. In an interview from November 2024, Sam Altman told Andrew Ross Sorkin, economics reporter for the New York Times, that the NYT He had positioned himself in the wrong corner of history. "We can talk about it and debate it, and we will—I think—in court," Sorkin replied.