Experimentating with Flux

The goal for the exercise was to analyse the output difference between how AI machine and human see data.

Using a same set of questions, it will be prompted to both the AI image generator & myself as the test subject. I will be drawing without reference just pure memory based.

Method

Only the first visual generated from the prompt will be used for this exercise.

Likewise for the physical drawing, there will be only one time and using a pen will prevent from any erasure.

The test was to engage the model with personal questions that I could have a comparison with the sketches I have, how perhaps how data works when it’s being prompted to the AI generator.

Having a rough idea of how AI training model works, it probably tried to categorize the text prompts into vector embeddings that matches the nearest/similar vector in the vector database and these are the results similar or nearest. I found it rather interesting where these are the first visuals from each questions, perhaps suggesting that text prompts that I input has the vector embedding values that was nearest to the images in the database.

I do realize that with subsequent same prompts, it probably changes the image but how closely it resembles the initial image might be something worthy to test? Giving it more complex questions is something that I’m thinking of trying with the Flux AI but the value of the test is something I need to consider.

"Non-Human Centered Designs" to "Raising awareness"

We have always been designing human centered designs but in near future we might be dealing with non-human centered design which leads to AI-centered design. Hence, what do we know about human-centered design and the difference to non-human centered design. More readings about non-human-centered design?