Reflections on a year of AI art
TensorArt’s first anniversary marks about the same length of time I have been involved in image generation. This, I think, is a good time to reflect on my thoughts about generative AI.When a colleague told me about Dall-E, I could hardly imagine anything more wonderous than the ability to conjure up an image that existed only in my head, with no more than a few well-chosen words.Through Dall-E I discovered ChatGPT, and I was equally amazed by a machine’s ability to create a coherent article out of a few rough ideas in my mind. And then the floodgates swung open, and suddenly AI seemed to be everywhere. In short order I became aware of Stable Diffusion and MidJourney, and I adopted the former as my go-to image generation AI model.Battling my way through a bewildering variety of SD implementations, mostly of questionable quality, chancing upon TensorArt (just shortly after the release of SDXL) opened a whole new world for me: access to variety of high-quality models that could generate images on par with more expensive rivals. I was instantly hooked, and months later, I still am.I can draw and paint reasonably well, but modesty forces me to admit that I am no great artist. I’ve done quite a few photorealistic vector illustrations in my day, but those would take up to 15 hours or more to complete. Having the ability to achieve that level of realism in minutes (and nowadays even seconds) remains mind-blowing.There are probably millions of people like me; people who have the vision to create beautiful art, but lack the manual skills to realise that vision. For people like us, generative AI is the Red Bull of creative endeavours – it gives us the wings our imaginations needs to soar to dizzying new heights. It allows us to express ourselves as we have never been able to do before, and to savour the indescribable satisfaction of producing a well-executed image that may otherwise never have seen the light of day.Generative AI, whether it be Stable Diffusion, ChatGPT, or any of the other implementations, is an incredibly powerful tool. I use it as a hobby, but also more and more as an integral part of my work. I am sure others do too. But with great power, they say, comes great responsibility.As much as AI has the potential to be one of humanity’s greatest inventions, it also has the potential to be one of the most destructive – not in a Terminator-type rebellion-of-the-machines kind of way, but as a threat to the very fabric of our society. I could delve deeper into this subject (and if interest justifies it, I may), but that’s outside of the scope of this article. Suffice it to say that, as with most human inventions, we need to find ways of using it to improve humanity rather than destroy it.As our species takes the first steps into this brave new future, seemingly brimming with promise and potential, we are at a junction where we can decide how to use this tool. I take heart at the fact that we in the TensorArt community, and other like it, seem to already have made our choice: as a universally accessible tool to give expression to the greatest gift we have: human imagination.
