Is This Ethical AI?

Large Language Models and Generative AI are, as far as I’m concerned, poison designed to kill our humanity. They are rotting our brains, destroying the environment, and bringing only questionable economic benefits (there are no social ones, as far as I can see).

I am so fervently, avidly anti-GenAI that I removed all Microsoft products from my standard workflow in favor of Ubuntu as an OS, LibreOffice for my document creation, and PCloud for my cloud storage, all because CoPilot kept infecting my software, trying to tell me what to do and constantly prompting me to use it when I DON’T WANT TO GODDAMMIT. One client has demanded that I use Office for their work, so I have one spare windows laptop that touches only their work and is quarantined from everything else, and I am using a dodgy lifetime Office license I bought off an auction site. No subscription.

I’m also seeking options to get away from WordPress.com because they keep shoving at me. Assholes.

I also abandoned Adobe Lightroom for my RAW photo development because it was not only hammering GenAI tools into everything, it was charging me more to do so.

At the same time, even I—diehard as I am—recognize that machine learning and analytic “AI” tools have their place. You want to use AI to improve cancer diagnosis and can PROVE that it’s helping? Hell yeah, go to it. And I’ll even concede that machine learning can be useful in less vital areas, as long as it is approached ethically and doesn’t encroach on what it means to be human.

Which is a big part of this, right? If AI is a tool, it must 1) do something right and 2) HELP people, rather than taking away from us. Tools do not steal our skills. Tools do not make it harder or more frustrating to do a job. Tools do not make us less capable as thinking, creative human beings.

So, when the free and open-source software that I now use to develop my RAW photos, DarkTable, announced that it was introducing AI-powered modules, I was deeply skeptical.

But reading their policy, I am perhaps now cautiously optimistic.

The full policy can be read here but the key points that make me think it might be OK are these:

1. It is only used for image correction: denoising, artifact removal without generating, color restoration etc.
2. All processing is on your machine, not at a data center
3. Object removal is only done using assets within the photo (like cloning) rather than generating things that weren’t there

These sound like tools. Things that I can use to polish and perfect the photos that I actually took, rather than creating things wholesale that were never there. And do it without engaging with the libertarian tech elite so intent on taking humans out of the equation.

At the same time, the models themselves worry me. Were they trained unethically? Am I benefiting from stolen creativity?

So, while the jury is still not quite decided, I feel like this could be a model for ethical “AI” tools in this very confusing modern age. But hey, maybe I’m wrong and it’s all just a burning hellscape of soul-destroying tech from top to bottom.

It’s hard to do what you think is right, maybe. But that’s why we have to keep trying to do it.