OK, so suppose I want to make 3D models in Blender which will become REAL objects, like with a METAL 3D printer - is it worth it trying to compete with AI in this regard? I’m talking of generative things like Autodesk Dreamcatcher, and whatever it’s Blender equivalent is (what is it?), and I ask this SPECIFICALLY considering lattices - a structure with a lattice in it is lighter than the same one without, and it is impossible for human beings to think of this type of thing - for something which will have to obey the laws of physics.
So - is it worth it trying to beat Dreamcatcher or other, possibly AI, software?
Find an AI which can answerit for you 
AI has the tendency to mimic biological designs. Which in turn leads to chaos-like or fractal-like patterns. As for example if you look at “bone structure microscope” you can get an idea of the shapes are almost the same idea as the AI.
Perhaps this is because cells have this exact level of sophistication for that exact level of scale they live in. And also these patterns are for internal structures only in order to achieve certain benefits such as strength-to-weight ratio. However taking that sort of shapes and scaling them up to the human usable level, it might make no difference.
Though there have been many tests, that show that A.I. fractal shapes are equally strong or stronger in some cases, no one seems to noticed a revolutionary trend in them to rewrite the entire manufacturing engineering books.
In terms of engineering, you need ease of maintenance, or predictability of patterns, definitely having parts that are simple to manufacture and easy to repair. On the other hand in bones of living organisms they self-healing or self-repairing (when possible) so this is not a problem.
There was a test also done. They let an AI design a CPU from scratch and again it made it like roots of a tree and such. Despite for example having better benefits in terms of voltage used (shorter distances means less wires and less voltage) this is a design and engineering hell. The risk is far too great to adopted so much unstructured design.
1 Like