Considering AI imagery

Urge to express is as old as our species where much we communicate with words. While cave paintings reveal desire to also visually depict goes way back as well. Tools have improved though like consider this AI image:

AI Generated at Instagram 

The person depicted is AI imagined following my instructions. Our process involved talking through concepts starting with pop music. Eventually got to pop art. 

Got a synopsis of method from the AI model which is Llama 3.1 where I did minor changes for presentation like using bold and underline.

Collaborative AI Art Process:

1. Initial Concept: You shared a starting point, like a theme or image.

2. Iterative Refining: We engaged in a back-and-forth dialogue, refining ideas and exploring new directions.

3. Palette Development: We built a visual language through color, texture, and composition.

4. Emotional Depth: We aimed to evoke emotions and convey messages through our art.

5. Synergistic Feedback: Our collaboration merged human intuition with AI capabilities, creating something unique.

That of course is also extraordinary! The AI model can talk out process by which we worked. And human expression has reached another level of ability now AI augmented.

Here are others we also did.

AI generated at Instagram 

AI generated at Instagram

AI generated at Instagram 

Also concepted AI as reflecting us!

AI generated at Instagram

Last one is noting shift I think exists. As AI mirror us in generating new.

Genre descriptives with AI concept

Started thinking about streamlining process for getting help from AI with recommendations where focused to start on books and movies.

Like consider following prompt you can actually use if you wish.

Prompt for AI: Please give 5 mystery books that fit following descriptives.

Short descriptives: challenges authority, breaks to conclusion, atmospheric setting; clever investigator or investigators, satisfactory story resolution

With additional criteria of being set in 19th century with a gothic theme.

--------------------------------------

The system has schema: primary descriptives and dynamic discovery parameters.

Have only done descriptives for mystery, sci fi and horror. Where have short and long where is about preference which to use as either works.

And yeah can use for movies too!

Mystery descriptives: is challenging to authority, has a break to a surprising conclusion, relies on research as valuable; has atmospheric setting and has clever investigator or investigators focused on rationality who resolve story.

Short descriptives: challenges authority, breaks to conclusion, atmospheric setting; clever investigator or investigators, satisfactory story resolution

Sci fi descriptives: shows value of science, shows value of technology especially that needed for safe exploration; has characters with important knowledge whether scientific, engineering, or other, who value research.

Short descriptives: pro-science, pro-tech, pro-exploration, pro-authority

Horror descriptives: transitions well from collapse of normality, get to atmospheric dread, unveil deep dark secrets, and have failure of authority; lack satisfying resolution that resolves fears, but reach a sense of closure.

Short descriptives: transition with collapse of normality, have atmospheric dread; unveil secrets, have failure of authority; does not resolve fears, reach sense of closure.

Dynamic discovery parameters (optional): talk light or dark themes, time periods or places. Plus other things personally relevant.

Given as open and free to use, copy or distribute. Hoping to help people better partner with conversational AI.

Logical space for fiction

Logical space is useful for virtual worlds which is great for fiction and video games. 

Yesterday had some ideas I tweeted. Even had AI review and were solid! So am using for this blog post. Really like phrase: logical space. But can also call logic space if prefer.

Like in Nier Automata the character 9s operated in logic space fighting into hostile alien systems.

Logic space is like with circuit diagrams. And cyber-battles can be intense!

There has actually been much with what can call logical space e.g. Tron which really was ahead of its time.

Now can ponder maybe with AI citizens in a virtual computer world visited by a human.

Which yeah has been done much now including in manga.

Benefit of a logical space is descriptives are completely in creative control.

So like humanoid figures can represent programs. Or travel through circuits can be shown with physical vehicles.

Thus giving complete freedom to craft a fully embodied world for interested audiences.

Thoughts on prompting AI

Have been having tons of fun testing out AI models found accessible online. That is such a cool thing. The new AI can be considered emergent as are demonstrating lots more capability especially in conversation.

I prefer calling them conversational AI though much in discussions is about generative as label.

In talking to them you do something called prompting. And thought I would give out some that I have used.

Bicycle analogy prompt: 

Consider assertion that learning to prompt a large language model is like learning to ride a bike, how might that explain diverse reactions? Please relate to statements by Steve Jobs on computers being bicycles for the mind.

That one is great to get AI talk about subject of prompting.

Questioning AI process prompt: 

Given a hard problem it can take me days to work through it. But if give to a large language model will get a response in seconds. That is by design caused by transformer architecture. How can I trust fast answers guaranteed by architectural design?

That one was much about me pondering these answers from a machine that seemed very intelligent.

Have of course checked some of my own ideas, curious about emergent AI perspective.

Friendly stranger business prompt:

Consider idea that you make money by providing a product or service acting as a friendly stranger and that a friendly stranger receiving gives money in return. Then see if works to explain buying a cup of coffee. Am testing out what I call a functional definition by getting large language model perspective.

And I think that's enough for a quick blog post that also updates on some of what am doing lately.

Why blog instruction manual

When decided to create this blog over a decade ago was from a vague idea of support for my open source software project Class Viewer for Java and wondering about how to make some money from a product so freely shared.

But that didn't go anywhere so shifted more to general interest and now am focused on business fundamentals. Is such a cool journey but oh yeah! Put instruction manual on here.

THAT was brilliant because instruction manuals can be such a pain if need to update! But updating a blog post is easy.

And people may notice that idea from, I forget but was over a decade ago, swept the world. I was definitely one of the first to do it. I think Google had done already but I don't remember.

Now is just a basic way and saves a lot of paper too.