Very cool to see. I wonder about the thermal performance. Titanium is typically used over aluminum in aerospace when relatively high temperatures are experienced, (but not as high as the engine) the leading edge of the wing in supersonic jets for example. Being composed of carbon, would it be susceptible to combustion, like diamonds are? Still, with the density of styrofoam, you could certainly replace aluminum in low temperature applications too!

If the structure is highly regular that makes me guess they would have good creep performance, although that's probably hard to test on such a small sample. It'll be interesting to see if and when the fabrication can be sped up from two days for a pea-sized sample.

[–] Lugh 2 points 2 days ago (1 children)

"For example, if you were to replace components made of titanium on a plane with this material, you would be looking at fuel savings of 80 liters per year for every kilogram of material you replace," adds Serles.

I'm impressed by two things here. That something so light could replace titanium, and that it was discovered by AI.

[–] just_another_person@lemmy.world 10 points 2 days ago (1 children)

It's not "discovered by AI". It's not a novel idea that a model spit out, rather, because that's not a thing.

They generate large permutations of already know data and combinations of things and have these models validate math and test a massive amount of variables very quickly. It's just quick sorting, essentially, which is all AI is really good for. Pretty sure they didn't want to get laughed at for aaying"AI", which is why they used "machine learning" in the title anyway.

[–] stetech@lemmy.world 1 points 1 day ago

Pretty much this. A lot of people nowadays conflate AI to mean LLMs, but this model almost certainly had no “language” part…