Skip to main content
Back to blog
ā€¢3 min read

Everyone's Making AI Art Now

AItechart

remember two months ago when i wrote about dall-e 2 and thought it was wild?

things have escalated.

what happened

stability ai released stable diffusion as open source. anyone can run it. on their own computer. for free.

meanwhile midjourney is taking over twitter. everyone's profile pic is suddenly a portrait of themselves as a fantasy character.

the floodgates are open.

my experiments

i spent a weekend playing with stable diffusion locally. observations:

  1. my laptop sounds like it's preparing for takeoff (GPU go brrr)
  2. the results are hit or miss
  3. prompt engineering is apparently a skill now
  4. some of the hits are genuinely beautiful

i made a prompt that was something like "atmospheric digital painting of a cozy programming workspace at night, rain on window, lofi aesthetic" and it produced exactly what i imagined.

that's the part that gets me. it's not just generating random art. it's translating what's in my head to an image. that's... new.

the debates

artists are mad (understandably)

  • their work was probably in the training data
  • the tools make art "too easy"
  • the copyright situation is a nightmare

tech bros are excited (predictably)

  • it's democratizing creation
  • it's a new tool, not a replacement
  • artists said photography would kill painting too

my take (tentatively)

  • both sides have valid points
  • this isn't going back in the box
  • we need to figure out ethics/copyright/attribution fast
  • being anti-technology doesn't work long-term, but neither does ignoring legitimate concerns

what this means for me

as a CS student interested in machine learning, i feel like i'm watching the future arrive in real-time.

six months ago, image generation was a research paper. now it's a tool i can run on my laptop in my bedroom.

what happens in the next six months? next year? next five years?

i don't know. but i want to understand it. not just use the tools, but understand how they work. why they work. what they can and can't do.

that feels important.

the bigger picture

AI art is cool and controversial. but it's also part of something bigger.

generative AI is advancing faster than anyone predicted. text, images, code... what comes next?

i'm equal parts excited and anxious. which is probably the appropriate response.

my stable diffusion setup generated 47 images last night. i went to bed at 3am. no regrets.