ArchiTECHt Daily: Strata news and insights into AI

Obviously, today's newsletter is publishing much later than normal, so I am going to keep this brief
ARCHITECHT
ArchiTECHt Daily: Strata news and insights into AI
By ARCHITECHT • Issue #37
Obviously, today’s newsletter is publishing much later than normal, so I am going to keep this brief and just highlight a couple of things worth reading.:
  1. This macro-analysis of the Intel-Mobileye deal from Stratechery. It’s a great look into the economics of chips to begin with, and also self-driving or otherwise-smart cars. The only thing it’s missing, if you ask me, is a reference to the data center, where Intel getting inside cars might also be a way to ensure Intel remains the chipmaker of choice on the backend of the connected car workflow, as well.
  2. This trio of announcements from quantum computer maker D-Wave Systems about (1) Volkswagen doing quantum computing; (2) partnering with the Virginia Tech / DoD; and (3) Google’s Quantum AI Lab upgrading to the latest D-Wave system. I’m always inclined to write off D-Wave because it’s been going for a while at there still hasn’t been a big commercial breakthrough that I can tell, but maybe perception is reality in this case and the promise that it’s close to one is enough to keep up the momentum.
That is all.

Sponsor: Datos IO
Around the web: Cloud and infrastructure
Companies like Facebook and Google could buy a lot of these things as they keep jacking up data center bandwidth.
This is a good interview about the challenges of building an open source project. Long story short: It cannot fund itself.
medium.com  •  Share
Basically, how does Google balance CIOs and software developers in its go-to-market strategy? This is the biggest question for pretty much everybody, but which AWS was able to largely avoid early on by just creating its own market.
This is a fair take on the latest release from my former employer, which includes container pods and 1-click deployment of even more software packages. Community is a bigger challenge than software.
Around the web: Artificial intelligence
No, this doesn’t mean Skynet either. But AI systems that don’t forget as easily means AI systems that can infer context and perhaps be practically useful in real-world situations.
It’s the guy who co-founded Geometric Intelligence, the company Uber acquired in December and that catalyzed Uber AI Labs. The former chief scientist, Gary Marcus, was the other co-founder, but he resigned last week.
Yeah, accountability is kind of a big deal in the military. “Because the computer said so” is not a very good reason to do something, or a very good defense.
The stakes aren’t as high as Google’s recently announced startup challenge, but building cooperative Minecraft players does sound like fun (and doesn’t involve starting a company).
Keras is a big deal among folks trying to build their own deep learning models, as is TensorFlow. But the billion-dollar question is who, if anyone, will monetize general-purpose AI software.
This is a smart, existential take from SwiftKey’s Ben Medlock on the question of whether AI can really be smart like humans. Basically, he argues, millions of years of evolution connecting our brains with our bodies will be hard to replicate in machines.
aeon.co  •  Share
Sponsor: Marshal.io
Around the web: All things data
Driven by banking and manufacturing. That’s a lot of dough, but spread over umpteen million vendors, SIs and consultants, it’s probably reasonable.
www.idc.com  •  Share
The Cloudera Data Science Workbench is one of the company’s first steps toward an actual application, targeting the folks who actually try to make us of systems like Hadoop and Spark. Hard to tell if there’s a SaaS option, though—which would be a good idea.
No one every accused MapR of not pushing the envelope in terms of product offerings. It’s new “mini-cluster” is a really smart idea and potentially a foot in the door to winning industrial IoT workloads.
A roundup of announcements out of Strata this week, all targeting the morass of products and projects that companies need to use to do anything useful with their big data.
Its idea to create a “data layer” for products, built using smart packaging, is solid. It starts with RFID, but there’s a lot of other information to capture about the lifecycle of a consumer product.
Last week, Google announced a cloud data-preparation service built using Trifacta’s software. Data prep is arguably one of the most important, and most overlooked, aspects of actually doing big data (and AI, for that matter) right.
Basically, Rheos provides lifecycle management and monitoring across eBay’s streaming pipeline and across data centers. The effort that goes into building something like this is why people think most companies will just adopt cloud services.
Did you enjoy this issue?
ARCHITECHT
The most interesting news, analysis, blog posts and research in cloud computing, artificial intelligence and software engineering. Delivered daily to your inbox. Curated by Derrick Harris. Check out the Architecht site at https://architecht.io
Carefully curated by ARCHITECHT with Revue. If you were forwarded this newsletter and you like it, you can subscribe here. If you don't want these updates anymore, please unsubscribe here.