Well the future is here . . . if you can explain what you want to create then a computer can now generate an image of it for you. If you want a “picture of Donald Trump riding a donkey while eating a hotdog” I guarantee you will be amazed at the results. I have been using DALL-E since it was first available to the public and I was pleasantly impressed at the improvements when DALL-E 2 came out. When it launched the images were good but not quite up to an amazing standard of work where you think people need to be worried about their jobs, but the landscape of AI has changed once again.
As is the case with AI we get to see massive exponential improvements over short periods of time and AI image generation has gone from a nice little toy with a few use cases to a monumental change that is going to reshape the future.
AI generative art has now reached a point in quality that you would happily accept it as a professional's work. What would have taken an artist hours to do can now be done in less than a minute.
There are now lots of different image-generative AI’s to choose from including DALL-E 2, Stable Diffusion, Jasper Art, Fotor GOArt but my go-to AI is Midjourney (for the time being). A lot of these AI systems use the LAION 5B data set that is made up of over 5 billion images with descriptions, it is a very large dataset taking up 380TB of storage.
When I sat down with my 9-year-old daughter to ask her to imagine any sort of artwork she wanted on her wall she said “a kitten jumping into water”. So we jumped into Midjourney and gave it “/imagine a kitten jumping into water on a beach with a sunset background” and we got these 8 images in under a minute.
If we turned the clock back only a couple of months, if I wanted artwork like this I would probably be paying north of $300, now I get it for free via an AI.
