For clarification, when I say “AI” throughout this post, I mean machine learning models such as Midjourney, DALL-E and the like.
Reasons Against using AI in my art:
I enjoy the thinking and craftsmanship involved in my art making
I love my art and my art loves me back
I have a problem with unsolicited art scraping
I have a problem with the carbon footprint of machine learning models
Illustration ©️ Kat J. Weiss / T-shirt mockup by Mockups-design.com
Where I might give the benefit of the doubt?
AI is very good at coming up with average solutions so it could be a way to gauge the average of what’s out there and innovate
As a solution to this problem: let’s face it, many customers — not all, but many — aren’t looking for something unique or bespoke. Most of the time they see your portfolio and commission a simple version of something you have already done before. Not necessarily a bad thing, sometimes basic is is sufficient, but it’s bad for your soul, I think, because as artists we’re bent on always innovating and creating the most super special thing ever (not exactly practical). Enter… your very own personalised Illustration Vending Machine! An eco-friendly machine learning model solely trained on your own database of work. Clients can use this to generate images in your style, for a fee and a package of usage rights — similar to how stock images work. And YOU my precious artist friend, can get back to seeking treasures in the depths of your creative diamond mine! Customers who want something more bespoke can still get in contact with you, and hopefully with some change in your pocket from the Illustration Vending Machine, you can be much more selective about who you work with. What I’m proposing is for workers to simultaneously take 21. century tech innovation and the fruits of their labour into their own hands.
To generate some parts of product mockups to showcase the application of my illustration. As you can see I’ve used a couple of free product mockups for my illustrations across my latest blog posts and newsletter. That got me self-reflecting on a couple of ethical aspects:
If AI-generated mockups are basically pasted together from existing product photography, then, according to my own value system, I must be opposed to that
I checked the source of all three mockups I used to see if they were AI-generated, but couldn’t tell for sure. Two of them are from mockups-design.com; according to their FAQ, they do not use machine learning models for creating mockups. Let’s hope that’s true!
My conclusion is that generating a product mockup with a “simple” or “basic” shape, against a background devoid of any creative direction (which also includes unique lighting situations, in my opinion) probably does not fall under copyright infringement
Having said all of that, I do not feel comfortable using AI-generated elements in my creative process, so I’ll be staying in my lane and observing developments. Perhaps there is more nuance to add to this, but I might need an entire Modern Art era and its developments figure this one out.
In light of recent events: RIP Suchir Balaji, former OpenAI researcher turned whistleblower who exposed how the company’s practices (allegedly) infringed on US copyright law