My early AI enabled experiments have mostly been image based, which is an easy leap for me to make given my illustration background. I've really enjoyed having silly ideas for imagery and then seeing what ChatGPT can create. This post collects some examples.
Adapting existing artworks
I've fed in various things I've made in the past, and asked ChatGPT to use these as a basis for creative images. This feels slighly uncomfortable given the understandable backlash from creatives towards AI and model-training, but also feels like the most exciting of all of these examples as it's the most unique.
Work related
Corporate life is full of presentations and pitches, and as somebody who likes to positively disrupt, it's been fun using AI to inject fresh imagery into various scenarios.
This was used in some team training to articulate what service design is - it took several attempts with ChatGPT but within a few minutes I had a visual that communicated something fairly complex.
This one was used in a presentation to pitch some ideas - I love pushing retro imagery into a work situation, and look forward to doing more of this.
Random
This street party invite shows one of my favourite parts of the ChatGPT image generation dance - type. This was an early draft and you can see where the copy is broken, but also kind of beautiful. I eventually fixed this using Figma which was a good early test of an AI to human workflow.
I'd cycled into work on my brompton and listening to a James Brown mix on soundcloud - all the prompt I needed.
Conclusion
There's no deep learning here - just a wally playing with a new toy. I'm excited to see where people take this, and will keep playing with new methods of creating imagery,