# Ramblings from a non-practitioning AI hater


Incoherent text about AI, what's wrong with it and how I'm not as consequent as I'd like to be. 

> This text was written without any use of AI.

### Things changed

I recently moved deparments but mostly kept my IT [DevOps](https://www.reddit.com/r/devops/comments/1f3dnik/how_do_you_explain_devops_to/) role. Before that, I've been on parental leave for half a year (best time ever), so I didn't really follow the industry. I knew some co-workers have been asking AI before and eventually so did I, but my attempts were 2 years ago when the quality was real bad. It didn't really matter at work. I've been following AI discourse on the Fediverse, which is very vocally anti AI. I am anti AI, at least in theory, and theoretically support most of the reasoning against it. But I also noticed AI creeping into so many Open Source projects lately and former naysayers starting to make use of it.

Coming back to work, I found that I joined the fucking slop brigade. Things changed. Most tasks are done with AI of any kind, even ones where I'm sure the people are capable of doing them faster manually. The documentation they gave me for Onboarding were mostly AI generated. My boss is vibe coding the Sloperator 9000 (some agent management tool to manage different AIs‽) in his time off. That's across techies and management, no matter the task. I hear similar things from everyone working somewhere IT and also more and more from people I know from different contexts, for so many different things. I got handwritten greeting cards where the text pattern seemed way too familiar, the takeaway across the street seems to be vibe-cooking according to their menu and even the local tattoo studio of all places can't be arsed to put human made art on their windows. My friends are asking AI how their new cabinet would look in each corner, at which point they should visit a doctor with their sick child and how they should to their tax return.

Thing is, I get them. This hurts to admit, but I **want** to use AI. It's so convenient. Google AI shows me what I'm looking for in most of the cases, I don't always want to look at the dictionary page, wiki entry, docs or spec sheet. Claude answers my endless technical questions about how things work when my head is in brainstorm/recherche mode in the middle of the night. What's the best setup for wireguard in this use case? What's the flaws in my plan to migrate my NAS? Yada yada. It's still the new glossy thing and it gets addictive real quick. It gives my brain what it wants in that moment.

### Why I hate it

Man, the externalized costs. Everyone is aware of them, most prominently the environmental impact, and yet, they are conveniently ignored. Some megacorp under fascist leadership leveling another forest for their mega-datacenter with an attached dedicated gas-powered plant, sipping all the drinking water? Bad I guess but whatever, my coworker needs to create a "skill" that automates five emails a year. Some people are downplaying the environmental damage or are even hoping that super-AI will find a way to deal with this (other than turning it off), but most have just accepted it and continue. What does one little funny prompt have to do with burning the planet? And of course, without context, that's right. But they are one of millions.

It's literally built on destroying the web. I had to implement countermeasures myself on my servers to stop the crawlers from [DOS](https://en.wikipedia.org/wiki/Denial-of-service_attack)ing them. Hardware prices skyrocketed out of affordable range. Efficiency gains get eaten by the [rebound effect](https://en.wikipedia.org/wiki/Rebound_effect_(conservation)), capitalism fuck yeah.

I don't even care about copyright. I do care a lot about re-propertisation, though. Internet Archive rules, offline copies are good, I'd love to have a copy of the public internet and some interface to just ask questions about the stuff! Make it all available for everyone. But this is not a neutral interface. It's a tool by a company that wants to make money off this. **Even if the knowledge pool would have been "neutral" and not already contain all the -isms, blind spots and biases that a largely white male internet has, the company will contiuously tune it to whatever fits their agenda.** One day it's influenced by an upcoming election where they favour a candidate, the other day it's just their ad partner and sometimes both. This sucks and will just get worse once they have more people hooked. And I'm not even talking about deskilling and other social effects or ever-growing complecity in code and elsewhere here.

I always think and speak of other people in this context but I'm no better. My little AI researches are in no way more justified than theirs. I know all of this and still I can't let go completely. It keeps my usage limited, that's all.

### Lack of conclusion

This would be a good point to wrap it up. Tell the reader there's not only back and white but a sensible middle ground. But honestly? There isn't. How many drained swamps for AI data centers are too many? One. Admittedly, I'm not sure where to go with this text. I've never written a blog article but this was spinning in my head and had to go somewhere. Maybe someone reads it or even replies? Even has similar feelings? Let me know.