Turnabout is fair play. Setting aside the overall ethical shitshow that AI presents, I don't really have much objection to turning their own tech against them.

LLMs are an incredible technology, and I would have loved to see them developed organically over time with sensible regulation and strong ethics....it would have taken many more years, but it also wouldn't have stolen the sum of human creativity, generated a hype bubble that threatens the global economy, wrecked the consumer hardware market, or been deployed in the most shortsighted, environmentally destructive ways imagineable.

But that's not what we got, so evoking the same fear and uncertainty that we must face in those who lead global megacorps is not out of bounds, as far as I'm concerned.

I've been threading the needle between my AI loving techbro indiehacker friends and my AI hating artist and FOSS geek lefty friends for the last year now and the dogmatism has been interesting. The extreme of it for AI lovers I don't really get in my friend group since I think it manifests mostly in AI psychosis meltdowns by tech CEOs like with the Proton Mail guy, or religious worship by the anyone with one degree of separation or less from anyone on the the AI doom cult temple sex priestess gang bang Sankey diagram.

For the AI haters the dogmatism turns AI usage basically into a moral contagion. I saw similar with the cancellation of Framework: https://blog.calebjay.com/posts/moral-contagion/#where-i-pus...

To them, even a small use of LLMs is a moral endorsement and material contribution to the pillaging of natural resources by data centers, the racial profiling and inherent biases baked into LLM output by their training corpus and prejudices of corporate leaders, the theft of artists, writers, and FOSS contributors for training data, and even could cause irreparable brain damage (they inevitably have studies to link to, on LLM usage reducing problem solving abilities or creativity). Also LLMs still can't write good code, to these people.

I'm having a hard time demonstrating the enormous activist potential these tools present, especially in their current token-subsidized form (Claude max still is a hilariously obviously unprofitable endeavor). IMO turning these tokens into liberationist software (OP project and similar) is basically a form of wealth redistribution. Moving beyond clean room software implementation, once you get to the possibly or frankly illegal things you can do with LLMs, their activist potentials seem pretty obvious to me. Cracking and reverse engineering proprietary software, developing hacks against The System (what if we leveraged LLMs to take out Russian datacenters that flood the internet with ultra right wing propaganda? Or... Industrial sabotage <.<), automating social media content in the same way the far right is doing (or the opposite - responding instantly to all far right fake news, solving the "it's easier to create it than debunk it" problem), creating agitprop, or hell creating floods of disinformation. I mean depending on someone's value system in terms of ideologically valid strategy, there's a lot of possibilities unlocked by LLMs!

Personally I'm interested in anything that simply takes the winds of profit out of the sails of as many things as possible, which is why I like projects like this one and why I spun up a similar one a few weeks back: https://github.com/508-dev/genairosity Though I'm looking at smaller niche things like, milk collection software or whatever.