AI may soon predict how electronics fail:
Think of them as master Lego builders, only at an atomic scale. Engineers at CU Boulder have taken a major step forward in combing advanced computer simulations with artificial intelligence to try to predict how electronics, like the transistors in your cell phone, will fail.
The new research was led by physicist and aerospace engineer Sanghamitra Neogi and appears this week in the journal npj Computational Materials.
In their latest study, Neogi and her colleagues mapped out the physics of small building blocks made up of atoms, then used machine learning techniques to estimate how larger structures created from those same building blocks might behave. It's a bit like looking at a single Lego brick to try to predict the strength of a much larger castle.
"We're trying to understand the physics of devices with billions of atoms," said Neogi, assistant professor in the Ann and H.J. Smead Department of Aerospace Engineering Sciences.
It's a pursuit that could be a boon for the electronics that underpin our daily lives, from smartphones and electric cars to emerging quantum computers. One day, Neogi said, engineers could use the team's methods to pinpoint in advance weak points in the design of electronic components.
The project is part of Neogi's larger focus on how the world of very small things, such as the wiggling of atoms, can help people build new and more efficient computers—even ones that take their inspiration from human brains. Artem Pimachev, a research associate in aerospace engineering at CU Boulder, is a coauthor of the new study.
"Rather than wait for years to figure out why devices fail, our methods can give us a priori knowledge on how a device is going to work before we even build it," Neogi said.
Journal Reference:
Artem K. Pimachev, Sanghamitra Neogi. First-principles prediction of electronic transport in fabricated semiconductor heterostructures via physics-aware machine learning [open], npj Computational Materials (DOI: 10.1038/s41524-021-00562-0)
(Score: 5, Touché) by Thexalon on Saturday June 26 2021, @09:20PM (4 children)
A major reason for electronics to fail is when they make an impact to a solid object moving at approximately 30 m/s. Usually as the result of maddeningly stupid proprietary UI's that tell me that something is wrong and I need to consult with my system administrator, but give no indications at all as to what that something wrong might be and no way for me the system administrator to get said indications.
"Think of how stupid the average person is. Then realize half of 'em are stupider than that." - George Carlin
(Score: -1, Troll) by Anonymous Coward on Saturday June 26 2021, @10:14PM
If the AI is any good, the only practical use for it will be to assist Jew-run companies in their quest to design products that fail after their warranties, but as close to the warranty period as possible. It is the M.O. of the Jew to create new markets where none previously existed, which is why Jew-run big pharma gets paid not only for their toxic vaccines, but also for the drugs used to treat vaccine complications. Jewish bureaucrats in Big Med also profit handsomely from the scam with $100,000 ER bills that get foisted onto the patients and taxpayers because big pharma is indemnified from the damages their vaccines cause.
(Score: 4, Funny) by Runaway1956 on Saturday June 26 2021, @11:04PM (1 child)
If congress would just pass a law, setting terminal velocity for electronic devices at 3 m/s, we would save the lives of countless electronic devices.
“I have become friends with many school shooters” - Tampon Tim Walz
(Score: 2) by Thexalon on Sunday June 27 2021, @03:26AM
They'd have to make sure the law applied to hammers too.
"Think of how stupid the average person is. Then realize half of 'em are stupider than that." - George Carlin
(Score: 3, Insightful) by DannyB on Monday June 28 2021, @04:31PM
The easiest way to predict when electronics will fail is to check if they are approaching the warranty expiration date.
The server will be down for replacement of vacuum tubes, belts, worn parts and lubrication of gears and bearings.
(Score: 2, Interesting) by Anonymous Coward on Saturday June 26 2021, @10:27PM
It's some clever pattern matching connected to a neural net, or something like that.
Using the "AI" buzzword just muddies up what ever cleverness is actually in play here.
It's not like chip makers are flying blind. For many years now they have learned how to grade chip life, based on data recorded at every step of the production process...and heavy duty multivariate statistics. The chips that go into mainframes (with years of uptime) may be the same design as in consumer goods, but the mainframe chips are known to be better made.
(Score: 1, Informative) by Anonymous Coward on Saturday June 26 2021, @11:33PM (3 children)
(Score: 2) by looorg on Sunday June 27 2021, @05:26AM (2 children)
That isn't really a "how" tho, it's more of a a "why" isn't it? It's sort of like saying that most of the electronics fail due to unexpected power surges or heat. Same things, it's now how they fail, it's a why they fail.
But it would seem like so many other things, they will fail via the path of least resistance, so the shittiest or cheapest or worst made component or part there of is probably the first to go and fail and then there is a cascade of failure from there.
Not sure if water is a bigger killer then say heat or some kind of power failure.
(Score: 0) by Anonymous Coward on Sunday June 27 2021, @01:31PM (1 child)
Once I drop my phone and the screen cracks I don't particularly care to know exactly how the screen cracked on a molecular level. I just care that now I am upset that my screen cracked. Yes, I can analyze what happened exactly and I don't even need an AI to do it. Who cares.
(Score: 2) by looorg on Monday June 28 2021, @07:49PM
Fine in those specific cases mentioned I might not perhaps see the point on a personal level either. Still one would hope this is used to create better chips, worst case better planned obsolescence.
But still it's interesting cause for a lot of components it's not the entire component that fails when it "fails". Some components just fail and it's gone but some IC as an example might not completely fail. It could be only part of it that has failed. Still it's failed and things will be borky at best after that but depending on what it does and how it does it there can be remedies. In some cases it might not even matter. I seem to recall back in ye' olden 8bit days that Clive Sinclair bought a lot of really cheap ICs he used for memory that had an abysmal failure rate in the higher allocations. But it didn't matter since he never used that part for memory so those legs were dead legs in more cases then one. It was cheap and that is all that mattered. I'm not suggesting that for today but still it can be interesting to note why and how a chip, or other circuits fail.
(Score: 0, Offtopic) by Anonymous Coward on Saturday June 26 2021, @11:41PM (2 children)
Why do gay dudes have an obsession with jock straps? I don’t get it.
(Score: 0) by Anonymous Coward on Sunday June 27 2021, @12:59AM (1 child)
Why are you obsessing about gays and jock straps?
(Score: 2) by coolgopher on Sunday June 27 2021, @03:24AM
Maybe he's just trying to understand himself better? Nothing wrong with that.
(Score: 3, Funny) by js290 on Sunday June 27 2021, @03:39AM (1 child)
electronics fail when you let the smoke out...
(Score: 2) by kazzie on Sunday June 27 2021, @07:21AM
Electronics: Smoke-in is good for your health.
(Score: 2) by crafoo on Sunday June 27 2021, @11:05AM (2 children)
This could improve the lab-life of electronics I guess? It really sounds like they are searching for a problem to solve with their admittedly interesting simulation software.
Anyway, electronics fail due to vibration, moisture, heat. all external environmental effects and all cause failure far sooner than rated lab life.
(Score: 3, Insightful) by maxwell demon on Sunday June 27 2021, @02:18PM (1 child)
In an ideal world, this would be used to improve the lifetime of electronics. In the real world, this more likely will be used to design well-timed failures for planned obsolescence.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 0) by Anonymous Coward on Sunday June 27 2021, @06:00PM
Exactly. They'll do a much better job of planning the failure to occur just after the warranty ends.
Good point, what financial incentive do they have to use this to keep the product working longer?
(Score: 1) by pTamok on Sunday June 27 2021, @10:13PM
Well, if I ever get a message like the following from an AI, I'll be mighty suspicious:
"I've just picked up a fault in the AE-35 Unit. It's going to go 100 percent failure within 72 hours."
(Score: 2) by Rich on Monday June 28 2021, @09:33AM
It sounds like they consider atoms moving around. On a related topic: The CMOS 4000 series (introduced as COS/MOS by RCA, btw, but that may have sounded a bit too communist back in the days, when it had to be AST/RO for the US...) had issues with electromigration. These ICs would just fail after a few years in operation. It seems the process was revised over time and this no longer happens. The line was sold from RCA to Harris around 1986, I think, and again to TI in 1998, which explains why TI offers SN74HC... and CD74HC... chips. Today, there are also the "AHC/T" lines. Does anyone know when the process changes happened and what die from what heritage really ends up in the chips we get now? (e.g. are the old RCA/Harris fabs still going, maybe for military qualification issues, or does TI stick their single AHC line into everything, because it will match all specs anyway?)
(Score: 2) by J_Darnley on Monday June 28 2021, @10:22AM
Will it be able to identify that you used cheap chinese-made knock-off capacitors that will swell, burst, and fail in a few years?