Models of computerization and automation have been developed for over 80 years. A certain sense of the healthy embarrassment led most people involved in these researches to avoid calling it "Artificial Intelligence." In keeping with the spirit of our times, the new techlords Elon Musk, Peter Thiel, Mark Zuckerberg, or Jeff Bezos have poured hundreds of millions into social media, academia, and the press to promote the "Artificial Intelligence" hype and normalize this expression. But their ideological project is not innovative.
AI refers mainly to text synthesis machines (and to a lesser extent, image and pattern analysis and classification machines for "autonomous" cars and deepfakes). These machines are incapable of producing new information; they don't "think" about what they're writing, using only the probability of what will be written next, according to the databases they've been programmed with. As such, there is no imminent awareness or a new entity that wants to destroy us like James Cameron's Terminator. The AI hype is essentially propaganda, the main purpose of which is to accelerate layoffs, to fuel financial speculation, and to divert investment and resources toward a new jump into the abyss by economic and political elites.
AI's main appeal for the general public isn't even the probabilities that build generally coherent texts and lists, but the language enhancement phase, a new coat of paint that produces an almost human language. They call it "Artificial Intelligence," but the real name is Large Language Model. The most famous models are ChatGPT, Claude, Gemini, DeepSeek and MechaHitler (Grok).
Considering the disastrous state of information on the internet today, language models are already suffering from a kind of Mad Cow Disease. Just as cows in the 1990s got sick from being fed bone meal and meat from other cows, so language models are degenerating as they are programmed based on data from the internet, where there is already so much data produced by other language models, in particular ChatGPT, that errors can swell to the point of incomprehensibility. Just as Mad Cow Disease contaminated humans, AI is definitely contaminating us.
The proposal is the same as always: Make the rich richer at the expense of those who work.
The promises that the techlords and politicians who have joined the AI hype have for us are generally false—the good as well as the bad. Language models are not going to end humanity or replace essential tasks in societies and do away with useless work. They are actually creating precarious, underpaid, and hidden work, among other things for the people who have to check that the answers given by the models are in polite language and not like Elon Musk's MechaHitler calling for Jewish genocides and mass rapes. This in no way means that there aren't already millions of people being fired in the hype that ChatGPT or another language model may eventually replace them. In many cases people are then rehired for less pay afterward.
Today's language models don't produce knowledge beyond what is already inside the databases that programmed them. We hear climate deniers claim that language models will solve the climate crisis, but this is redundant. Models based on scientific texts and decades of climate negotiations know how to solve the climate crisis, which has been public knowledge for decades—we need to end the fossil industry in the very short term so solve the climate crisis. Anything else is the ideology of collapse that has been embraced by capitalists. Models based on pseudoscience and random content taken from the internet will spew out garbage in response to any prompt. If what goes into the programming of the models is bad, what comes out can only be bad. So, the point is not that an AI will become too intelligent and wipes us out, the point is that there is no intelligence involved.
But this hasn't prevented language models from being used widely, with unknown and private algorithms, managing huge amounts of data. It is guaranteed that there will be misinterpretations of data and requests that will cause irreparable damage (in health, in criminal data, in energy systems, in the allocation of social benefits, as has already happened in several countries). There will be no one to blame for the consequences, as the billionaires who spread AI outsource their responsibility for all this with the backing of the political elites.
The dissemination of language models on a large scale corresponds to an ideological project of the technolords, selling the idea that humans are just organic versions of computers, reduced strictly to what they can produce. In capitalism, the main promise of AI is the abstract possibility of making a series of jobs redundant or unnecessary. The point is not even to make them redundant or unnecessary, but simply to create the illusion that they can open the door to laying off millions, without even having to prove how AI would replace those people. It's the eternal return to "increasing productivity," replacing labor with technology in theory. To install this ideological project on a large scale, widespread data theft and the end of privacy would have to be normalized, with surveillance systems and permanent punishment for the poorest. This has nothing to do with a great technological advance or any global awareness nonsense. The proposal is the same as always: Make the rich richer at the expense of those who work.
The scale of the ideological project based on "Artificial Intelligence" is catastrophic: firing hundreds of millions of people working in health, education, justice, science, art, public services, and the press with the vague promise of automation to substitute their work. This ideological project would also entail a massive expansion of data centers and network infrastructure, skyrocketing energy and material needs in the midst of the climate crisis. Techlords and the deluded politicians who support them care little if AI language models fail to replace most of the jobs they have already started to destroy. The techlords' doctors will still be people, just like their teachers, lawyers, and information services. For most of the world's population, what could be expected from such a project is more poverty and an incomparable degradation of any public and private services, handed over to automated parrots built with databases contaminated by other automated parrots.
===
Joao Camargo is a climate activist in grassroots movement Climaximo in Portugal and in the Climate Jobs campaign. He's an environmental engineer and climate change researcher at the University of Lisbon and the author of two books: Climate Change Combat Manual (in Portugal and Spain) and Portugal in Flames - How to rescue the forests.
Spread the word