Against “artificial intelligence”

2024/09/02 Landabidea Urresti, Xabier - Aisia eta Giza Potentzialean doktorea. Deustuko Unibertsitateko Euskal Gaien Institutuko ikertzaile elkartua. UEUko Komunikazio sailburua. Iturria: Elhuyar aldizkaria

• Adimen artifiziala

Unless the last few years are hidden in a cave, we all know what artificial intelligence is, that everything is changing and that it is inevitable. In a short time we have built a great hype of consensus between enthusiastic screams to capture those “technologies” as soon as possible and join their “revolution”. The only and best option is to give in to that force, as inevitable as gravity and death.

If the situation were different, I would write a column of opinion with more nuances, valuing possibilities and risks, combining criticism and praise ... But I would argue that, in the situation we are in, we have to be absolutely critical of the representation of artificial intelligence, and I will propose five reasons for this:

1.La artificial intelligence doesn't exist. What's called yes is, but that's mythology, not materiality. This is the label that we have applied to the different technologies, techniques, practices and disciplines (which are of the most interesting and real), but today, AA, like the ziritionea of the Volskwagen cars, is a huge marketing operation that does not describe reality, that it wants to create.

2 AA is a robbery. We have already witnessed the largest session of looting and privatization of power and symbolic capital in history. It’s not just another gold rush (although the NVIDIA Corporation is enriched by selling piquitos and digital shovels), this is the biggest theft in history, in which big immortal corporations have devoured our data and (it says fast) the entire human culture. It is a “crime of piracy” of the SGAE that makes those of us who remember the campaigns in the relationship between corsairs and crowns think.

3. AA is not a tool (for our objectives). The expansion of data storage and computing capabilities has opened up new possibilities for applying probabilistic models in very different fields, from machine translation to learning that elephants are called by their own name. But the ones we've called AA are not neutral instruments. Its aim is to expolite the whole natural and symbolic environment and to eradicate surplus value. If they're hammers, we're nails.

4. AA is that brother-in-law of Christmas dinner. You know it doesn't calm down. More than anyone knows what it is and offers simple solutions to all the ills of the world. He knows nothing, but he has full confidence in his abilities. Google will tell you that Rome was founded in 1870 and ChatGPT that the fastest marine animal is the big hawk, and they will tell you with confidence. Surely that's what we needed, a wave of synthetic sister-in-law that will poison the entire Internet.

5. AA is false. It's not just that as text and image generators have no sense of meaning or referent (they operate probabillistically on simple indicators), it's not just an automatic brother-in-law. As an opaque box, we are promised mystery and magic, even revelation and salvation, in a religious way, and we are continually told that they are technologies impossible to understand. Like any teleological coming, of course, it has its messiah, its prophets and its sects. But after all, we're pushing superstition, not knowledge.

Is there something interesting about what we've called deep studies and machine studies? Of course. But in general, I would say that the situation is very worrying and that they have led us to board the train so that we do not have time to ask where we want to go or if we want to go.

 

-> Some reflections on artificial intelligence: challenges and consequences