Can companies reach peak intelligence without making the climb?

It says Issue #101, but this actually is the 100th issue. So, hooray! And thanks for reading!A couple
ARCHITECHT
Can companies reach peak intelligence without making the climb?
By ARCHITECHT • Issue #101
It says Issue #101, but this actually is the 100th issue. So, hooray! And thanks for reading!
A couple of articles caught my attention on Thursday that suggest, to me, if very indirectly, that we might be getting ahead of ourselves talking about the short-term effects of artificial intelligence. It’s not because AI isn’t powerful or because there aren’t perfectly aligned applications for it right now—there absolutely are—but, rather, because many companies still have not mastered using data as a tool for competitive advantage and product feature.
Before I go further, here are the two items I’m talking about:
The NYT article points out how some car companies are going to extreme, ranging from chip fabrication to quantum computing, in order to gain an edge in building smarter cars. The HBR article points out how we’ve evolved from Hadoop and R hosted locally to APIs, edge computing and machine learning. 
“Analytics in the past were created for human decision makers, who considered the output and made the final decision,” the HBR article states. “But machine learning technologies can take the next step and actually make the decision or adopt the recommended action.” Based on most of what I’ve read and heard in the past several months, it actually doesn’t appear that most enterprises are anywhere near this level of machine learning adoption, much less automation. Maybe in the warehouse, but certainly not in the boardroom. 
While cars are mechanically smarter, the driving experience—the way people interact with their cars—is, save for perhaps Teslas and some other high-end vehicles, more or less as dumb as it’s ever been. Forget autonomous driving: simple things like WiFi, real-time traffic info and working voice controls are still out of reach in many vehicles.
Companies like Google and Facebook are leading the way in AI in part because they long ago mastered big data in terms of infrastructure, collection, analytics and as the foundation of new features. It’s institutional knowledge as much as it’s technical knowledge.
Companies and institutions of all types absolutely should be looking at where and how they can apply technologies like AI (and, I guess, quantum computing) today. We see successful applications all the time, but often for specific tasks that are relatively easy wins (I’m thinking things like document review in law, or optimizing factory robots). I’m just not certain we can reach peak automation and intelligence in our economy without many companies actually making the climb. 

Sponsor: Bonsai
Sponsor: Bonsai
Artificial intelligence
This is probably a good idea, but it’s a little disappointing if we’ve been getting scanned for years and agents still can’t identify weapons (or at least suspicious items) with meaningful accuracy.
I, like others,  spoke a lot about Andrej Karpathy’s computer vision prowess, but this post notes his expertise in other areas that could also be a boon to Tesla. It’s maybe worth noting, however, that he has never led a product-focused division like this. At least that I can tell.
Technological implications aside, you have to wonder how this would affect startups like Rigetti or other large vendors like IBM. Being first isn’t everything, but (if the world is ready) it certainly helps.
It’s a small, Canadian company with a big dream of connected cars and smarter ways of regularly assessing risk. But its current revenue model—playing policy broker to consumers—highlights how far it has to go.
It’s reassuring to see more experts suggest that human intelligence can’t be decoupled from our brains and bodies. That doesn’t mean AI isn’t intelligent, but that it’s fundamentally different.
This paper presents a chicken-or-egg problem when it comes to analyzing medical data, which can be plagued by errors, but also just uncertainty on everything from diagnosis to cause.
arxiv.org  •  Share
This is interesting if you follow advances in reinforcement learning, as it presents an easier and computationally cheaper approach to transferring knowledge between agents.
arxiv.org  •  Share
Sponsor: Cloudera
Sponsor: Cloudera
Cloud and infrastructure
This is a hugely important issue for cloud computing providers as well as anybody offering a web platform, and Google is not alone in pushing for chance. In a nutshell, Google is asking for some basic agreements on which countries can ask for what kind of data, and under what circumstances.
blog.google  •  Share
This is one of those areas where smaller, specialized cloud providers can probably carve out a niche. The question is what happens if they’re successful enough that the big boys catch on.
This seems like a valuable service, especially for companies whose IT staff are not well-versed in the container world. Don’t sleep on IBM in this space—it’s an active open source contributor with a big-time sales org.
This is a good post about a question lots of people (including me) have asked. Things get complicated, of course, when vendors and come into the picture with a zero-sum mentality.
redmonk.com  •  Share
I met Nell at GeekWire’s cloud conference earlier this month, where she gave a talk on DevOps. She’s really smart and really nice, and here’s her advice on everything from managing meetings to keeping rabbits from wrecking your rig (or getting fried).
The lengths Netflix goes to in the name of optimizing user experience are truly impressive. Every company thinking about streaming should pay attention.
medium.com  •  Share
This post covers a lot of microservices ground and tackles a lot of reasons why companies do the things they do, even if they maybe should not. But a big, important, point of discussion is how to think about databases.
Sponsor: DigitalOcean
Sponsor: DigitalOcean
All things data
This is really smart from a product perspective, IMHO. You could use Elastic’s technology to build something for APM, or you could just buy it from them.
It’s a real use case for partners and companies with valuable data to sell, and a good way to gin up some more sales because both parties must be Snowflake customers.
There are so many database options out there, most making a valid reason for why they exist. TimescaleDB is no exception, especially in the age of IoT.
Sponsor: CircleCI
Sponsor: CircleCI
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.