The DDoS attack started in 2018, though at the time, I didn't see it coming. I remember standing with colleagues in a hallway, at the Center for Mind/Brain Sciences of the University of Trento. The conversation went like this:

- Have you seen that new overblown model that performs so well on the X and Y benchmarks?
- Well, given the size, they might as well use random projections, would probably work great.
- Haha! Yeah. Wanna grab a coffee?

The model was the Transformer. /1

Fast-forward seven years. Billions of dollars have been spent on the Transformer and co. The EU has just announced a 200 billion € initiative to promote #AI, following other governments across the globe. Millions of people, crushed by the demands of 'efficiency', now think that a machine writes email better and faster than they do. And wow, the new tech can even fire people (in all senses of the term.) Still wanna grab a coffee? /2

In the last seven years, I have had to teach Transformers, reply to questions from over-enthusiastic students and concerned family members, write countless lines (publicly and privately) about the many issues associated with 'AI'. Hell, I even had to build the thing from scratch to illustrate its problems. I am on stage every month with a tiny Transformer which you can unpack, showing to an interested audience why the thing is all hot air.

Problem is: it is not my job. /3

Don't get me wrong, I do enjoy the outreach. I even think that looking inside a machine helps people understand better why they are people and not machines. Language is one of these things that are never taught in school (kids know more about the Big Bang than they do about the faculty that helps them think, believe and communicate). So perhaps showing how #AI fails to do language is the best way to describe what language actually is. But it is only a small part of my job description. /4

I am computational linguist. Wanna know what I was working on before 2018? I was doing computer simulations of (parts of) the human semantic faculty. I was showing how knowledge, beliefs, real and hypothetical worlds live inside the human mind, in multidimensional spaces. I was showing their geometry and manipulating conceptual shapes the way you move concrete objects in space. And it was beautiful and it had so much potential to explain aspects of our humanity. /5

How much real science have I done in the last 7 years? A fraction of what I wanted to do. My field was engulfed by #AI hype. Literally. You can read the reports highlighting how industrial involvement in our scientific community drastically increased after 2018. I got out of academia because of it. But then I couldn't get away from the fact that the general discourse had also been hijacked and that some reaction was needed. /6

I am an introvert. I don't enjoy talking, fighting, arguing. I don't actually like social media (sorry Mastodon, you're great nevertheless). I like being alone with my eyes closed, wondering what would happen to this or that conceptual structure when you rotate it through space. This is what I'm actually good at, or at least what I feel comfortable doing. But you all know the sign: "It is so bad that even introverts are here." /8