Martin Hähnel

Work, Values, AI

Published:
Updated:
Tags:

I found this post pretty interesting: My Work, My Values, and AI.

To put it simply, everything that comes after the first section is just obvious to me. It's obvious to everyone I speak to, even folks who use LLMs regularly, under duress or even voluntarily. Literally everyone knows all of these things. And yet, here we are. So leading with them is more likely to be interpreted as a justification along tribal lines than a choice of considered conviction, and that adds nothing to the conversation.

If you're pro-LLM and read this far, did I change your mind? What about if you are anti-LLM? If you use it because your employer forces you to, and you know that you're just training your replacement anyway, did this make you feel better or worse about your situation?

I'm willing to bet that the answer is "no" to all of those, in every case.

Right. Everyone knows about the harms associated with AI and no one is convinced in either direction to do anything different. I argued something similar recently:

I think Doctorow, Tante and me we all oppose purism, we just all diverge on what follows from this: Doctorow seems to think reappropriating technology is enough, Tante believes in critique and I don't believe in either. I think we are shaped more by society and its structures, the systems that shaped and still shape us. If we are able to change anything it's incremental, minuscule and less grandiose than what these thinkers seem to believe.

The article that sparked me wanting to write this down, has the goal to clarify that their content is produced by them. LLMs are sometimes used on the input side, but output is untainted.

I think it's good to leave a paper trail or line in the sand for yourself and anyone wanting to know what your stance is or was about AI. Because something seemingly simple as a few posts about a contested topic can really clarify your own thinking. So in the end posts like this are for ourselves. They make us say out loud what we are thinking and in so doing make our opinions observable. I am happy that I don't have a platform, as I don't believe testing opinions is the right way to go about figuring out what one's personal ethics are. That process is a deeply personal, but I am also happy that I have a blog, because if I make up some bs or do something awful (hopefully not too awful!) I am much more likely to change course if I blogged publicly about it.

Only you can decide if a certain thing fits your values. Or is at least not diametrically opposed to your values. My simple test for AI recently has been to observe my hesitation to continue doing something:

This is as personal as it gets: Do I actually feel hesitation that warrants a closer examination? The article speaks of that, too:

[T]he conversation tends to end with the author having a sad from the push-back [(this is about right-wing tech bros)], and harvesting likes of their public pouts rather than exhibiting any interest in understanding further.

Understanding further, not in the sense of accumulating facts[1], but to introspect more, is something I find very valuable. I hope - I don't dare to be too certain - that this will prevent me from becoming a person who claims neutrality but is actually just a right-winger.[2]

Anyway, I'm less interested in the political overton window problem, but I am interested in a more generalized version of the "overton window problem of the self": In that, it seems to me, only self examination, with a paper trail (i.e. in public) has any chance to catch shifts of acceptable behaviors that make it possible to ask questions like:

Questions like this can be answered on a surface level, but if you let them, a continuous practice of self examination can be transformational. It is similar I think to "meditation but with thoughts": To not get distracted by the simple answer, the pleasing answer, the well-flowing argument, the cute sentence, or whatever, is similar to get back to the breath. And I think this is what "understanding further" implies to me.

P.S.: And as always, nowadays (not that I'm complaining, really), I don't have really time to make this post any better than it stands now after basically one-shoting it: Maybe that's a good thing, but it certainly is not as round as it could be.

P.P.S.: Argh. No time to put this/integrate this better right now, but I want to be honest: I used AI (Mistral) to jog my memory (historian of science here) about Popper's points in the post I am talking about. Also, it is very likely that I end up paying and using Mistral for my personal use going forward, which will definitely warrant its own blog post at some point soon.

  1. Or examining well-known arguments like Popper's sidestepping of the paradox of tollerance, as is done here. ↩︎

  2. A great post about that is The Sports Bar, although the main point is slightly different. I think I am a left-leaning social democrat. ↩︎