Sam Sartor
# My Objection to Generative AI2026-6-8

A Bit of Context

I have conflicted feelings about AI. I spent the last 5 years of my life working on diffusion models, and I’m proud of what I’ve done (see my dissertation). I’ve always tried to do well by the artists my work is intended to support: training only on CC0-licensed data, releasing everything open-source and open-weight, avoiding projects where slop is the main application. But some days I hate having to tell people what it is I work on.

I hear people say that “AI is bad because it steals from artists”. Or “AI just doesn’t work well”. Or that “AI is destroying the planet”. I don’t agree with the specifics of any of those arguments, but do understand where people are coming from. The vibes are rotten, the feelings are melancholy, and we are all looking for a convincing reason “why”.

This post is not a rigorous. Just an attempt to dress my own feelings in a bit of “why”.

A painting is a series of decisions. You start with a blank canvas and then you decide where to place the brush, and decide again, and again, and if you make enough decisions you wind up with art.

My objection to generative AI is that it exchanges those decisions for a series of random chances. That isn’t hyperbole, it is the rigorous mathematical definition of a generative model. Generating an image takes away all those decisions, the intentionality, the context, the meaning, leaving only the pixels themselves.

And maybe you think the pixels are all that matter. But then why is the Mona Lisa worth more than a JPG of the Mona Lisa? Why go to an art museum at all? Why ever sit down and watch a movie when you could scroll? Why not just inject yourself with heroin? It is the same dopamine either way. Humans need meaning, need context, need intentionality, we need to make decisions. The world is already facing a crisis of meaninglessness; I’d rather not make it worse.

And it isn’t that I object to generative AI simply existing. It should be there when you want to make a quick joke, fix a bug, or generate a bit of wallpaper. I use Claude a lot, and nano-banana a bit, it all seems like good fun! But all around me AI is being sold to replace human thought, marketed that way, pitched that way, designed that way. I can’t help but feel insulted. Respect and kindness are in short supply, but you can find a trillion-dollar IPO on any street corner.

In my free time I love to play Scottish folk music. That strikes some people in the UK as odd, because I’m American. But a lot of Scottish traditions are only preserved today in North America. The old Scottish clans were evicted from the highlands, sent off to Nova Scotia, and replaced with sheep. Not because sheep are better at playing the bagpipes but because sheep are a better investment. Running around on the highlands waving swords and steaming haggis was a perfectly decent living for thousands of years, but it wasn’t optimal, and so it wasn’t allowed.

Maybe AI art will eventually look as good as human art, or maybe it won’t. Maybe AI code will eventually be as reliable as human code, or maybe it won’t. Either way we humans are already looking like the bad investment, like a poor use of money, suboptimal. Why invest in bridges, trains, or farms that are only needed by humans? Buy GPUs! Why invest in meaning, context, or intentionality if only humans want it? Sell slop! What is the point of making happiness when you could be making money?