this post was submitted on 12 Apr 2025
30 points (94.1% liked)
Ask Lemmygrad
942 readers
90 users here now
A place to ask questions of Lemmygrad's best and brightest
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I would argue that generated images that are indistinguishable from human art would require an AI use disclosure. The difference between computer-generated images and human art is that computers do not know why they draw what they draw. Meanwhile, every decision made by a human artist is intentional. There is where I draw the line. Computer-generated images don't have intricate meaning, human-created art often does.
I don't really see how a human curating an image generated by AI is fundamentally different from a photographer capturing an interesting scene. In both cases, the skill is in being able to identify an image that's interesting in some way. I see AI as simply a tool that an artist can use to convey meaning to others. Whether the image is generated by AI or any other method, what ultimately matters is that it conveys something to the viewer. If a particular image evokes an emotion or an idea, then I don't think it matters how it was produced. We also often don't know what the artist was thinking when they created an image, and often end up projecting our own ideas onto it that may have nothing to do with the original meaning the artist intended.
I'd further argue that the fact that it is very easy to produce a high fidelity images with AI makes it that much more difficult to actually make something that's genuinely interesting or appealing. When generative models first appeared, everybody was really impressed with being able to make good looking pictures from a prompt. Then people quickly got bored because all these images end up looking very generic. Now that the novelty is gone, it's actually tricky to make an AI generated image that isn't boring. It's kind of a similar phenomenon that we saw happen with computer game graphics. Up to a certain point people were impressed by graphics becoming more realistic, but eventually it just stopped being important.
Kind of unrelated but if you are to start to learn about AI today, how would you do it regarding helping with programming (generating images too as side objective) ?
Having checking the news for quite sometimes, I see AI is here to stay, not as something super amazing but a useful tool. So i guess it's time to adapt or be left behind.
For programming, I find DeepSeek works pretty well. You can kind of treat it like personalized StackOverflow. If you have a beefy enough machine you can run models locally. For text based LLMs, ollama is the easiest way to run them and you can connect a frontend to it, there even plugins for vscode like continue that can work with a local model. For image generation, stable-diffusion-webui is pretty straight forward, comfyui has a bit of a learning curve, but is far more flexible.
Thank you, I'll check them out.
the weird perspective in my work isn't an artistic choice, i just suck at perspective lol
Yes but you intentionally suck, otherwise you would just train for thousands more hours. Or be born with more talent. /s