ArchiTECHt Daily: Health care is AI's chance to shine

The field of artificial intelligence is in a funny place. On the one hand, it's hot: the term is all
ArchiTECHt Daily: Health care is AI's chance to shine
By ARCHITECHT • Issue #4
The field of artificial intelligence is in a funny place. On the one hand, it’s hot: the term is all over headlines right now and seemingly every other consumer application claims to be powered by AI. Just yesterday, I saw news about everything from marketing software to family-friendly WiFi routers touting their deep learning algorithms.
On the other hand, all this attention is bound to leave many users underwhelmed, if not disappointed. Many of the applications only arguably use anything resembling real AI, some are bound to not work well, and others are just uninspired. A refrigerator could perfectly analyze my tastes by analyzing the food in it, but I already know I like hot sauce. To quote a certain elected official—"Boring.“ 
However, one area where AI shows a lot of promise to really change lives is in health care, and not just via physicians consulting Dr. Watson. We’re talking about directly impactful stuff like diagnosing diseases to more behind the scenes stuff, like predicting how soon a patient will return to the hospital. AI is a great tool for augmenting human capabilities in areas where machines can do it better, and the field of medicine is rife with the types of complex relationships and latent patterns that machines are perfectly suited to recognize.
Keep reading for a round of very impressive AI news items from yesterday alone.

Wednesday in the world of AI + medicine
A study from Stanford, published in Nature, proves what many have long expected to happen: deep learning algorithms (in this case, ones originally developed by Google for the ImageNet computer vision contest) can learn what skin cancer looks like and diagnose it as accurately as a trained dermatologist. Imagine the resources—and lives—this could save if commercialized properly. (To read the whole paper, click on the link to it in this article.)
I remember speaking with Co-founder and CTO Ankur Teredesai about this research in 2014, and I’m glad to see it has found a home with KenSci. It’s not designed to diagnose disease, but rather to look at holistic data and predict who might get sick and when, and to optimize treatment plans for both cost and recovery.
And as the previous link evidences, there’s a lot to learn from analyzing data about who’s coming to the hospital for what, as well as how long they’re staying and other non-medical information. Kudos to the VA for sharing its data, which could be a big help to veterans in the short term, and possibly to a lot more people as a solid base of training data.
Around the web: Artificial intelligence
In case you didn’t know, Canada is a hotbed for AI research, startups and even corporate investment. NextAI, funded by several Canadian banks, is offering startups up to $200,000 as well as learning from leading researchers and technical assistance from Google, IBM and Nvidia.
If Gartner is publishing reports on something, you know it has hit the mainstream. 
Google is surveying Raspberry Pi users to hear what types of software tools they’d like to see, and deep learning seems like a winner. In a world where $100 maker kits do state-of-the-art object recognition … 
This is a good writeup of current advances in reinforcement learning (think Google’s AlphaGo system and Microsoft’s recent Maluuba acquisition) and where the field is headed.
Around the web: Cloud and infrastructure
No, I hadn’t heard of Faction either. And, yes, the investment involved both equity and debt. At this point, it’s safe to assume there’s a market for this type of enterprise-focused public cloud, but it’s not an Amazon killer by a longshot.
This is the first recorded instance of “We’ve seen this before with OpenStack” followed by a positive statement. Kubernetes is obviously a big deal, but it will not become ubiquitous this year. The “container wars” are jus getting started.  •  Share
Around the web: All things data
As usual, Tony Baer provides a smart analysis of the big data market, but maybe gives too much credit to CenturyLink. If you believe in data gravity, then the only data services that matter are running on AWS, Microsoft or Google.
The long story, short is to think seriously about what data you actually need to collect and where makes the most sense to process it. Really practical advice considering the potentially limitless volume of sensor data.
This list runs the gamut from Docker to TensorFlow, but there’s a whole of of big data winners, including Databricks, Apache Spark, BigQuery and DeepSQL, among others.
Briefly: do it. There are some potentially interesting findings here, as well, although your mileage will obviously vary.
Around the web: Security
Another company doing machine learning to detect threats. Someone should focus on learning users’ bad habits and preventing them from executing.  •  Share
This is a very descriptive account of Google’s KVM security practices, as well as vulnerabilities to specific attacks. Google has done a great job targeting specific folks lately (e.g., security and ops) in order to win them over on its cloud.
Analysis from shows a seemingly large talent gap in Israel, where there appears to be far more jobs than job-seekers. You’d think that would be the case everywhere given, well, everything.
Did you enjoy this issue?
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
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.