On AI Art

As many of you are aware some of the recent AI tools have been causing some noise. As it's common on the internet, there's no middle ground and a lot of it completely misses the point.

I'm in fact ambivalent about AI art. Can relate to the concerns. Can see the potential.

This is a collection of thoughts on the topic:

  1. If I had been working for decades to paint like Beksinski, I'd be pretty annoyed right now - that said, Beksinski already painted like Beksinski - and that's the only thing these tools can do at the moment.
  2. I'm also of the unpopular opinion that tools do matter. Working with pencils and working with oils is not the same and naturally leads to different results. Same digitally. 
  3. Some of these tools are... boring. The medium is the message. If I wanted to write things to communicate my ideas, I'd be a writer and not a visual artist. 
  4. I suspect most of us find it easier to communicate with images rather than words - the idea of putting a word barrier between us and the image completely defeats the purpose of creating images in the first place.
  5. There's also a mechanical component to it - the flow state is usually dependent on agency and (physical) action - a key element in sport/art - not sure how these tools can replicate that. 
  6. There's also the issue of how these systems are built - they rely on huge chunks of data to figure out the patterns - unfortunately, the good stuff is always what's outside these patterns - which may lead to a regression to the mean - an overabundance of hacks and mediocrity.
  7. My engineering background forces me to take Murphy's law seriously and consider the scenario of AGI: the program puts a Da Vinci out every second. There's the possibility that we simply don't care - like we don't care about computers being able to play chess better than us. 
  8. There's also the possibility that the AI version of myself is better, faster, and cheaper than I am, there's clearly a market incentive to go for the AI - as I take a few hours to make a character, not a fraction of a second. The issue of copyright becomes extra tricky too. 
  9. Finally and most importantly of all - we collectively need to start asking if something is a good idea in the first place. The future isn't always bright. Recommended reading: Cat's Cradle.
  10. I see the idea AI tool similar to NVIDIA Canvas, the artist retains full control of what matters and the AI takes charge of the boring stuff. I want to paint one tree, but 10? Please AI, take care of that.  

In Sum, 

  • Good AI: gets rid of all the boring stuff, without destroying the process and the artist's input.
  • Bad AI: will be able to create anything better than anyone ever will. Everyone wants the AI stuff. Humans can't compete. T-800 comes from the future to kill Sarah Connor. Artists continue to do their stuff and need to find a day job to pay their bills - nothing really changed.

An AI picture

Hope you've found this useful. If you agree/disagree/have something to add please reach out - this is a topic that demands discussion from all of us. I truly believe AI can make or break the world - let's hope it stays somewhere in between.  

Report