Martin Hähnel

Reverse Hesitation (About My AI Abstinence)

Published:
Tags:

So "the split" for using AI at work (necessary, therefore OK) and personal life (ethically questionable, but the desire to do stuff with it is great) has a new wrinkle:

1.) I'm happy that Mistral has a new Model out. If this is good enough for small coding stuff and a little chat - which it might be - I might test it out. Mistral is no innocent party and just as bad as the others when it comes to ethical AI, regulation lobbying and so forth, but they are a EU company. So voice becomes an option, especially because I am an EU citizen. So I could maybe justify using Mistral's services as long as I don't just loyally use it without speaking up.

2.) I notice "the pull" of AI on me. I am - maybe "of course" - hesitant to call it an addiction, but working with it ~40h a week makes its absence felt. Maybe that's why 1.) is interesting and seems justifiable - and in practice it isn’t - or at least might not be. I also think of the plastics metaphor @JulianOliver made (https://mastodon.social/@JulianOliver/117390234976335897): "[…]just as it's very hard to live without plastic, it's already very difficult to avoid slop in free software[…] Either we keep going & push for culture shift, or dust out a cave & wave goodbye" I am certain they didn't mean it this way, but to me the question is therefore "can we make the plastic better/more sustainable?" I feel like you have to engage with the material/thing in question to answer that.

3.) About that addiction thing: It is an extremely crude metaphor, not least because there's no such thing as just a tool (https://deadsimpletech.com/blog/no-such-thing-as-just-a-tool), but: I have become used to having this tool available to me for all kinds of tasks and I feel like because I do care about the output and ruminate ethical and epistemic implications, that my personal use - that is admittedly the simpler question - is alright. This is prefaced by believing AI is more of a hammer than a gun. My hesitation about having stopped using these things in my personal life (as opposed to somebody like @brettcodes who made the decision and ran with it) stem from a deeper hesitation about the soundness of the argument(s) that these things are all trash, these things can never be ethical and can only ever be made by proto-fascist tech bros. They are made, right now, by these people or people who don't concern themselves too much with moral conundrums, I guess (people who sit on tables with nazis and "don't mind it", maybe), but that's not, I believe the whole possibility space. They are used by all kinds of people - some more concerned than others. But definitely not only right-wing idiots. And part of me thinks that some (not all) of the AI criticism (mostly the quick fire takes on Mastodon/Bluesky and the like) is also plain nostalgia. Structurally, LLM tools like coding harnesses or chatbots have certain properties that mean that you gotta use them a certain, skilled way, I guess. The Zuhandenheit of LLMs and what follows from it means carefulness, skillfulness is mandatory. Like with a knife, maybe.

I don't know…