Changing Course

So im currently I’m in the process of completely switching gears in relation to my topic. I am now focusing mainly on AI with a current goal of looking at how it can benefit architects efficiency when it comes to iterating and filtering out non design related work.

These are my current sources that I’ve been referencing to learn more about the field of AI in design:

I am also using software such as chatgpt and midjourney to experiment and see some of the limitations in imaging.  I’m curious as to how these two software’s can be exchange information and how I can take a answer from chatgpt into midjourney and create a image solution to my question. I’ve already been testing some of these software’s to see the potential in efficiency in learning technical information. The example below shows a prompt I put into ChatGpt and answer it gave was surprising. It honestly gave me a better summary of how curtain wall connections work than most websites. I also tried this with rhino commands to see if it would talk about the animal rhino but it instead gave me a comprehensive list of instructions on how to use the command. So far I’ve been really impressed as to how the AI is able to answer my prompt in way I expect it to. Ive already been experimenting with midjourney in trying to get it to give me different pattern variations of a façade I originally created but need to look more into how to get clarity within the pattern I’m giving it.

In regards to my sources I’ve fully looked through one of them titled “creative ai, augmenting design potency”. I found it to be pretty helpful in understanding the differences between computational ways of designing. In the reading it talks about expert systems vs learning systems. Expert systems are more related to things like parametric design and requires a lot of human control/agency. Learning systems are supposed to be more closely linked to a neural network that filters raw data. The picture below is a good example I found of it. Learning systems are more closely linked to ai software such as midjourney.

The Reading argues about the creativity involved in each system. Expert systems are really creatively limited to the parameters in which they fit, while learning systems have a large possible domain of outcomes. He sees learning systems as having the ability to expand our creative domain. The idea that we can use AI as a generator of new solutions outside of the samples given is pretty interesting and novel. It can allow us to test ideas we have never thought or images we don’t have the time to image. But is this really increasing our creative domain in the correct way? In reality this learning system is taking huge sums of data from all over and is extrapolating information from that to create a result but this learning system isn’t really experiencing information the same way we do. We create our perception of things through past experiences and learning systems create perceptions of things through massive sums of data, building vocabulary images and understanding of those images. I guess my question is whether or not learning systems have the capability to understand things the way we do through our sense and whether this plays a role in the output of information presented by the AI. Since we operate on different levels of perception is there a level of creativity that AI simply cant reach?

Leave a comment

Design a site like this with WordPress.com
Get started