I've rarely used AI creatively. Never in music, except of course the algorithmic music generator I programmed in 2014, Oldfield 1, subsequently releasing an album of such tracks in 2015; thus I can claim that Art By Machine was the first commercial AI music album (algorithmic music was created far earlier, perhaps going back to the 1950s, but it's uncertain if this music was published commercially by a record label). Today, the term 'AI' generally means generative AI using neural networks; so 'AI' generally now means tools and output created since winter 2022 when Chat GPT appeared. Still, the Oldfield example already shows that AI usage isn't necessarily a simple yes or no. Also, as time moves on, everything is tainted by AI. Perhaps writers who have ever read some AI text and have become inspired or influenced by it could be said to be using AI, even if they, ahem, never have!
I've never used AI in writing, incidentally, not even for proof reading. In visual art I've not used it to generate digital art. I dislike the whole idea of this, just as I dislike the whole idea of music generation with apps like Suno. It's fine for people without imagination or those with no skills, but I have both. I also like physical media. I became an oil painter partly as a rebellion from the digital realm. I like CDs, I like oil painting, and rarely make digital art at all... but I do sometimes, and nowadays primarily for 'graphics'; that is my album covers, for book covers. Using generative AI in this context is, for me, more acceptable. This isn't 'fine art'. Fine art should be about expressing human values to other humans, and it should show personal skill and brilliance - this I value hugely. Making art should be difficult. One reason I dislike AI is that it's easy; any idiot can use it.
Yet, AI can be used creatively, to enhance and inspire, to improve imagination. It can also speed up a process which could be done mechanically and slowly.
I've not ever tried using AI for video, though I enjoy watching a few of these by Lanny Quarles, Peter Gric.
Before today, the only time I've used generative AI in art is image enhancement, attempting to improve poor quality reference images, and then only twice. This isn't always useful, as unrealistic artifacts seem to be created, but it can be of some use. I can filter those bits out; I know anatomy and how details like ears and hair should look even if the AI doesn't.
Today though, I decided to experimentally use AI to create reference images for my next painting, 'Punishment Of The Sun'. I have many ways of working, but amassing source references images for textures and scene elements is something I do for almost all paintings, and using AI for this rather than trawling free image sites like Pexels, or making my own, is a helpful and time-saving tool. One downside is that the resolution and quality is (ironically) rather poor, and there are those unrealistic parts; but those things aren't important for reference images I'll be painting from. To some extent, I like these flaws, as it requires more skill; I'm trained in improving images 'by hand' as I draw and paint.
With AI I can feed it my idea and work on generating a scene in a new-virtual way. In the past I've made 3D computer model scenes, and perhaps now favour making and lighting clay or cardboard models, but AI is another tool that I can use. It must be used with great care, not to compromise skill or imagination, but enhance both. The uniqueness of each artist is a vital spark that must be preserved; but even in AI, artists like Lanny Quarles have this in spades.
