Asimov's robot brains when?
The unusual properties of spin glass enable a form of AI that can recognize objects from partial images much like the brain does and show promise for low-power computing, among other intriguing capabilities.
Spin glasses are a way to think about material structure mathematically. Being free, for the first time, to tweak the interaction within these systems using electron-beam lithography makes it possible to represent a variety of computing problems in spin-glass networks
At the intersection of engineered materials and computation, spin-glass systems are a type of disordered system of nanomagnets arising from random interactions and competition between two types of magnetic order in the material. They exhibit “frustration,” meaning that they don’t settle into a uniformly ordered configuration when their temperature drops, and they possess distinct thermodynamic and dynamic traits that can be harnessed for computing applications.
AI algorithms developed in spin glass would be “messier” than traditional algorithms, Saccone said, but also more flexible for some AI applications.
Unrelated to the findings but my brain is trained to read anything after the word ‘paradigm’ as people high on their own supply, innate reaction
Also if this sees public adoption I will do literally anything to acquire a physical manifestation of the bogo sort algorithm
I had to look up spin glasses, they seem very cool. If people actually manage to scale these up to large systems successfully, it could be a big deal. But they'll need a lot of work to catch up to semiconductors. Physically constructing and measuring a weird new material of course, a good use of time, and the science is worth it regardless of whatever practical uses it has.
Okay, now that I've said good things about the science... wow, everything in this report makes my bullshit detectors scream.
-
"A form of AI that can recognize objects from partial images," refers to a Hopfield network; a form of AI that's not terribly useful and fell out of favor in the 80s.
-
"Low-power computing" - all of this is happening at supercooled temperatures. Even if there's no power draw in the circuit, keeping something at 30 K or whatever is not low power.
-
"allowing algorithms to be directly printed as physical hardware" - we can already do this in semiconductor land, it's called an ASIC.
-
so its like computer hardware made from magnets? man i wish i knew techno-geomancy.
