Skip to content

[FBT] on Sellouts and Savants

“Writing — writing of any sort, not just an email — is a thousands-year-old technological hack to get around not being in person; but that doesn’t mean that it is impersonal.” Audrey Watters

Maybe I’m a sell out

I gave Anthropic twenty dollars and I feel gross about it. Sorry and ashamed. I debated about telling you. Before I gave them my twenty dollars, I tinkered with other models. Everything local. Open source LLMs. This is the first time I’ve ever spent money on a technology and then felt shame.

Anger, shame, fear, disgust – negative emotions tell us something about our values. I’m now 100%, fully uninterested in the made up dichotomy between “rational” and “emotional”. Number A, they are not opposites. And Letter 2, the idea that one (rationality) is more important than the other (emotionality) is colonialist patriarchy’s capture of the human experience. Those dominant rationalists would have us believe that other ways of knowing, our intuition, sensory experience or, yes, you guessed it, feelings, are less valuable. The patriarchy says we should repress our silly little emotions, that’s not how the world works, etc etc. I say fuck right off to all of that, you patronising motherfuckers.

Street art by Hera

It can be acceptable or even admirable to have an emotional reaction and make decisions based on it with or without fully understanding the rationale. So I have been feeling a lot of curiosity, lately, despite knowing that my values and literally everything about the situational reality surrounding AI – from the unethical companies that are pushing it to the way it’s being used to murder civilians to the ecological destruction it is encouraging to the slop that is breaking my beloved internet – in no way aligns.

Over a year ago, when we wrote a policy document for Friends of the Earth, we worked with environmental activists and technologists to articulate principles that could help guide us in better decision making around AI. The wording is nuanced, and I, having helped to write these, helped to reword them again and again, want to adhere to them:

  1. Curiosity around AI creates opportunities for better choices.
  2. Transparency around usage, data and algorithms builds trust.
  3. Holding tech companies and governments accountable leads to responsible action.
  4. Including diverse voices strengthens decision making around AI.
  5. Sustainability in AI systems helps reduce environmental impact and protect natural ecosystems.
  6. Community collaboration in AI is key to planetary resilience.
  7. Advocating with an intersectional approach supports humane AI.

I don’t use AI to write (or to plan something written) because I write to think and if I’m not thinking, what’s the point? I don’t like reading AI generated stuff, and I certainly don’t consent to cleaning it up. I don’t run my writing through AI for “feedback” because I don’t want that kind of feedback. Statistical mediocrity is not my thing. If YOU, reader with a brain, want to give feedback, I’d be happy to hear it. If you don’t like my writing because of how I do it, why are you reading it? Kindly jog on.

Every once in a while, I use local models to correct my written German. If I’m writing something in German, it’s usually a letter to some sort of Amt (government office). Even after almost 20 years I cannot “Der, Die, Das” (neither can most Germans, so whatevs). I used to ask friends and family to make sure my German is ok, now I don’t. This is a net positive in our relationships. I don’t use AI to generate images. I did once, three years ago.

I have been feeling curiosity around the frontier models because I haven’t found  the other models I’ve tinkered with particularly useful. I am curious about what it feels like to make things in a different way. I used Claude cowork this week, and found it to be an interesting tool.

Maybe I’m a savant

I have a lot of jumbled thoughts about what we lose and what we gain with our ability to make interactive webby things without thinking through them in the way that we used to. Many moons ago when I was building webby things, I was learning about the guts of the internet. Now I know a lot, A LOT, about technology.

Let me be clear: AI assisted coding is fucking awesome for making people feel empowered to just make something they would never have been able to make without that help.

Is the code “production level”? Who the fuck cares, it’s their personal project, fuck off with your gatekeeping.

Does it work? Cool. Did they remember to ask the AI to check for security issues? Even better.

They’re not trying to create the next Google. But even if they were: do you even know how Google started? It was jank. Pomax

This week, I “vibecoded” for the first time. I now hate the term “vibecoding” even more than I already did. My thing is not finished, I hit my Claude limit twice starting to build an idea I had in 2018. An art piece that has lived forgotten in a local spreadsheet. It’s an idea that I recently referred to in the hopes of inspiring a foundation. It worked, they took it and ran with it.

Something that began as a rant about the glory of defiance and is now, after all these years, an unfinished homage to my feeling brain. After my limit was hit, I spent multiple hours working on my piece, something I haven’t even looked at in a long, long time.

I hit flow. I know the steps involved in interactive project development, I’ve been doing it for a quarter of a century. Using AI changes that process in no small way, but it would be useless without me knowing what I know.

While using AI is a different way to do things, I do not think it is always a better way to do things. I began to think about the fact that there is no objective definition of art. What is the objective definition of AI slop? When people say “AI cannot make art”, what do they consider AI art?

“I would like to suggest that slowing the fuck down is the way to go. Give yourself time to think about what you’re actually building and why. Give yourself an opportunity to say, fuck no, we don’t need this. Set yourself limits on how much code you let the clanker generate per day, in line with your ability to actually review the code.” Mario Zechner

Maybe I need help?

I need to go outside, that’s what I need. How are you?

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.