MIT dropped its list of 10 Breakthrough Technologies 2018, highlighting the advancements that could influence industries from manufacturing to enterprise to financial technology.
Don’t write off 3-D printing just yet. Sure, the technology has been around for a while and has added some spice to runway shows with fanciful, futuristic fashions and far-fetched footwear. But it has relied almost exclusively on plastic, which is fine for some applications but certainly lacking for others.
Now, however, it’s becoming cheaper and easier to print things like metals, and this advancement could prove significant. Think of how much time some companies wait on replacement parts for machines—and now, robots—in manufacturing and other sectors. 3-D printing could potentially change all of that. The biggest upside is less downtime, which in turn drives productivity and profits. 3-D-printing metal objects also could mean that companies could hold less inventory and create products on demand.
What’s more, 3-D printing could yield metals that outperform conventionally produced ones. According to MIT, Lawrence Livermore Laboratory, a major research center, developed last year a technique for printing a stainless steel that’s two times stronger than its conventional counterpart.
The first sub-$100,000 3-D metal printer launched last year.
All AI, all the time
Artificial intelligence (AI) will trickle down from the purview of mostly tech companies into other industries, according to MIT. Though some sectors of industry regard AI as cost-prohibitive and difficult to deploy, cloud-based machine-learning tools are changing all of that. An Amazon Web Services subsidiary offers cloud AI, and Google’s open-source TensorFlow AI library can be used to develop new machine-learning tools. Perhaps more importantly: Microsoft, through its AI cloud platform, Azure, is partnering with Amazon to build an open-source deep-learning resource called Gluon.
“Gluon is supposed to make building neural nets—a key technology in AI that crudely mimics how the human brain learns—as easy as building a smartphone app,” wrote David Rotman in the MIT Breakthrough article.
Related Stories
With each passing day, AI is getting better at discerning and creating. The ability to “imagine” is what—for now—separates the creator from the created. But AI is closing that gap, thanks to a University of Montreal Ph.D. student who decided to have two neural systems square off against each other in what’s called a generative adversarial network (GAN).
Here’s the point of the exercise: one system was deemed the “generator” and was responsible for making pictures similar to what it already had seen; by contrast, its opposite, the “discriminator,” had to figure out if the images presented by the generator were real or fake. Both systems were trained on the same data set, ensuring an even playing field.
The learnings from that experiment showed that the generator progressively gets so good at “faking it” that the discriminator can’t discern the real from the made-up. It’s an important step toward AI gaining the capability to ideate.
GAN made headlines in the past year when Nvidia, a chip maker, fed the system with myriad celebrity photographs and the neural network in turn churned out a number of incredibly realistic-looking pictures of (non-existent) people.
It may not be long yet before AI is fully capable of creating on its own.
Earth-friendly energy
Is clean energy from natural gas possible? Net Power, which operates a pilot power plant in the Houston area, is betting on it. The company is refining a technology that could one, generate power at the affordable cost of standard natural gas and two, contain all of the carbon dioxide emitted in the process. If it pans out, this clean-energy natural gas could be a more reliable than renewable energies while bypassing the high capital costs of nuclear.
Perfecting privacy on the web
The acronym zk-SNARK may be one to keep on the radar as it could be the key to web surfers proving they’re over 18 online without having to hand over personal information or transacting without having to share bank details. It stands for “zero-knowledge succinct non-interactive argument of knowledge.” (Yes, that’s a mouthful).
Cryptocurrency provider Zcash developed this form of zero-knowledge proofing, a cryptographic tool that’s generating interest alongside the rise of digital currencies. zk-SNARK allows users to maintain their anonymity while transacting, a departure from how blockchain-powered interactions typically work.
In a vote of confidence, J.P. Morgan added zk-SNARKs to its own blockchain-powered payment system in 2017. Still, the tool may not be quite ready for primetime; it’s slow, computation-heavy and open to certain vulnerabilities. But researchers in the field are hard at work trying to solve these issues.