The AI Con
- Victor Hugo Germano
- 7 hours ago
- 6 min read
This was originally created in Brazilian Portuguese
It’s hard to research AI in a skeptical way, without getting lost in the cacophony of dream sellers. Over the past year, on this blog, I’ve dedicated myself to presenting some of my research, and sharing my perceptions about these systems, the market that surrounds these solutions, and the countless controversies that have arisen from unchecked accelerationism .
Whether in the technical discussion about how these systems fail to deliver what they promise , or in the harmful consequences of the production system that enables the lie of automation and artificial intelligence, I have found more and more inspiring voices. And the book The AI Con , by researchers Prof. Emily Bender and Dr. Alex Hanna , brings an important critical perspective, especially at this time when bigtech CEOs are competing to see who has the most blatant statement about the future of technology:
"The danger [of AI] emerges from rampant financial speculation, the degradation of informational trust and environments, the normalization of data theft and exploitation, and the data harmonization systems that punish the people who have the least power in our society by tracking them through pervasive policing systems. But the Doomer/Boosters would have us looking the other way from all these real harms, bedazzled by their dystopian/utopian visions." - The AI Con
The AI Con , How to Fight the Bigtech's Hype and Create the Future We Want , is a book that reads like a podcast conversation. But a really good one! Packed with examples and direct references, the authors don't hold back on criticism and jokes, amid clear expositions of the use of automated systems with that impact the lives of people who already suffer from systems that don't work, or that persecute them.
The chapter that describes how LLMs work and tells a bit of the history of how the technology was developed and evolved in recent years is a great example of scientific dissemination at its best: you don't need to be a technical expert to know what's going on, and most importantly, you have complete autonomy in choosing not to use these systems. LLMs are just Synthetic Text Generation Machines , and we can interpret their output without imagining a mind behind the prompt trying to communicate with you.
The big problem with using these systems is not that they potentially allow us to do more with less: it's that we do less , with more potential for harm. And only when a large enough group of people have already suffered greatly from the consequences of these systems do we eventually discuss their impact and responsabilities. No matter how much we fantasize about future AGI, we have enough problems to solve today that these systems are only making worse, but also with no evidence that they will solve them in the future.
The book deals with real problems, affecting the lives of real people right now, and challenges the current narrative of technological abundance and utopian future that the market and media have embraced. And it makes a clear distinction from what has become a marketing cliché: Artificial Intelligence is a term used to hype up a set of technologies, not all of which actually work.
Bender and Hanna have been producing the podcast Mystery AI Hype Theater 3000 for quite some time now, which aims to separate the Hype from the reality, what is factually true in the market, and what is marketing for financial speculation. The best part of the podcast (and the book) is the strategy of using humor and ridicule to criticize what startup executives have been trying to push to the general public for a few years now: extraordinary claims require extraordinary evidence, and since no AI startup is capable of doing that, there is nothing better than making fun of the nonsense spoken by these companies. Most companies do not deserve to be taken seriously!
The book does this masterfully, and makes reading even more interesting! Using Ridicule as praxis, a phenomenal strategy!
My list of necessary books understand AI just grew:
You need to read this book.
Artificial Intelligence is neither Inevitable nor Mysterious
There is an aura of mystery surrounding everything that is said about AI in the media, increasing the feeling that we are on the verge of an explosion of intelligence and the eventual discovery of new forms of life: but it is all just delusional intent, or speculation propaganda.
While we are preoccupied with the dreams of AGI and science fiction speculations, we are increasingly faced with movements that seek the degradation of working conditions, climate catastrophe and political impact of information systems. In our eagerness to believe in a benevolent deus ex machina future, we forget the real consequences of these tools today.
I believe that Generative AI is the perfect tool for the social media machine :
Creating generic and quick content to generate any engagement, which is favored by the Social Media algorithm
Social networks rely on engagement to sell views to companies that pay for advertising
Algorithms rely on constant content to maintain relevance, driving platform usage
LLMs invade our information systems like an oil spill , and contaminate everything and put our own online social organizations at risk.
Our platforms feed on fast and average content, expanding our society of spectacle
Generative AI only accelerates what Cory Doctorow called the " Bostification " of digital platforms. And that could lead to the eventual end of the internet - but that's a conversation for another time.
More important than discussions about AI Safety and Alignment, automated systems are already causing real harm: in violation of civil rights , deterioration of working conditions , facial surveillance of marginalized communities , and state violence . These are real civil rights and human rights issues, not fake scenarios to play out in tech meetups.
"Modern logistics and information economies are built on automation, surveillance, and the reduction of people to mere objects, cogs in a machine. AI is part of a long tradition of global industry finding ways to replace labor, and is forcing degrading working conditions in the name of productivity." - The AI Con
I agree with the authors when they explain that LLMs were created to simulate the way people use language, to take advantage of a human characteristic: our ability to anthropomorphize objects, and our tendency to attribute meaning and imagine an intelligent mind behind everything we read. Believing that machines can be conscious is the first step to being manipulated by a computer program that spits out words in order to sound like a person.
With the fact that the output of LLMs is intentionally created to look like a subject interacting with users, it is almost impossible not to fall into the Subjectivity of machines, even though the truth is that Large Language Models are just presenting an extensive combination of groups of words that have a greater probability of appearing for a given context.
Add to this an entire industry to promote tampered benchmarks , research that is impossible to reproduce and a set of publications without criteria , and it becomes difficult to escape. And the examples of the use of LLMs for science, such as Meta's Galactica fiasco, also fail to hide an increasingly apparent reality: these systems are doomed to failure. No datacenter size will solve this situation.
This is an essential book to delve deeper into this increasingly impactful topic, and not in a good direction. The authors also suggest some influential readings on this topic:
The TESCREAL bundle: Eugenics and the promise of utopia through artificial general intelligence by Timnit Gebru and Émile Torres
More Everything Forever by Adam Becker
Great White Robot God by David Golumbia.
"A characteristic of our current hype cycle is that the con men are taking a series of tropes from science fiction—of artificial minds hell-bent on turning us into paper clips or Terminators waging wars for their right to exist (and to look cool on motorcycles)—and injecting them into discussions at the highest echelons of business and government." The AI Con
It is very comforting to find more and more authors questioning and criticizing the madness that our market has become: more material to study and more arguments to positively influence the future.
It's impressive how the book, which is recent, could already be filled with dozens of other cases that happened in the last few months. CEOs of billion-dollar companies, who used humans and sold solutions like AI , or companies that had to rehire professionals after ridiculously failing with the use of agents. Or more concerning: tech executives influencing the results of LLMs to push fake news and extreme political views.
We continue to seek a skeptical view of this madness.