AI Generated Images, 2024

The "AI bubble" that we're currently experiencing started on 30th November 2022 (a Wednesday for those asking), the day OpenAI released ChatGPT (Chat Generative Pre-trained Transformer) and everyone and their dog went bonkers for it, AI, and the new world order.

As I write this I am yet to experience anything astounding in my life from any "AI" from any tech company of any size. I have noticed that Google Assistant is getting worse, Google Maps is slower, but apart from that (which I assume is "AI" related), nothing.

Note: I am yet to work with any organisation that utilises "AI" in anything more than, "Ooh, look, shiny new toy".

Generative "AI" is actually what ChatGPT, Gemini, and many other types of "AI" are, systems capable of generating text, images, videos, code, or other data using generative models, often in response to prompts.

I am not impressed by text or code, I should be but rarely am.

I will always have a sense of wonder when it's used to create images or videos.

At the moment the systems out there such as DALL-E, Gemini, Stable Diffusion, and Midjourney all do the job with defects. These are rapidly getting better and soon it won't matter which one you use as they'll all do the job perfectly, and that job is creating ANY image your imagination cares to conjure.

Video is still early days ... for now. This will change very quickly.

As an example of not getting it quite correct go to Google's Gemini and use the prompt,

create a white image

It just can't do it ... but it will, and likely it's already too late and it's working now.

I am collecting a stack of random prompted images created by Gemini to document the increasing quality of that particular image generation system.