Tesla hires an AI expert, Walmart vs. Amazon heats up, and a quick lesson from Andrew Ng

Apologies in advance for this issue being light on analysis in the links section, but as I noted in t
ARCHITECHT
Tesla hires an AI expert, Walmart vs. Amazon heats up, and a quick lesson from Andrew Ng
By ARCHITECHT • Issue #100
Apologies in advance for this issue being light on analysis in the links section, but as I noted in the last issue I am traveling, which tends to throw off my publishing schedule. And I would rather publish late and light than not at all.
I’m actually in San Francisco for a few more hours, where last night I interviewed Andrew Ng in front of a large crowd of students at the Internapalooza event. Being on the field at AT&T Park was cool, especially for an average-at-best little leaguer some 25 years ago, but hearing Andrew guide students on how to think about a career in AI was also eye-opening. The one highlight I’ll share is his take on how startups or small companies can compete with large companies like Google in artificial intelligence: It all comes down to data. 
Essentially, Ng said, it’s of course really difficult to compete with large companies in areas where they already have mountains of data and will only generate more. You don’t want to compete with Google on applying AI to search (or probably videos or images), and you don’t want to compete with Facebook on applying AI to social media. But there’s a whole world of opportunity in areas ranging from health care to agriculture to education to enterprise IT where these companies don’t yet have the market cornered, and probably (with the exception of health care, perhaps) might never try. The trick is figuring out how to get access to the data you need, which, Ng noted, might take some creativity—or perhaps just reaching out to an organization that has it and working out a deal.
I have to say, I’m happy to see this data discussion getting more attention recently, because I think paying too close of attention to the latest-and-greatest models might be a red herring if you’re trying to do more than learn the techniques and tinker. Actually starting a successful business around AI means finding enough of the right data to train those models, too. Possibly first.
At any rate, I’ve written about a couple of times lately with regard to Facebook and Pinterest, if you want to take a look. AI training data is also the subject of the latest ARCHITECHT Show podcast, which you can find a link to below.
And here are three news items (well, two news items and one feature) from Wednesday that you don’t want to miss:
Tesla hires deep learning expert Andrej Karpathy to lead Autopilot vision (TechCrunch): Karpathy comes from OpenAI, and has done some great research and written some great blog posts on deep learning. Hopefully that continues at Tesla. He’s the second (at least) prominent researcher to leave OpenAI for the corporate world, after Ian Goodfellow returned to Google. (On a related note, Karpathy joins Tesla as Chris Lattner, who created the Swift language at Apple, left his post as VP of Autpilot.)
Inside Microsoft’s AI comeback (WIRED): This is a good read on what Microsoft is up to in AI, including its burgeoning relationship with expert Yoshua Bengio. I believe Microsoft is uniquely positioned to capitalize on delivering AI to the enterprise, which makes its investments in companies like Bonsai, and Bengio’s new enterprise-focused startup Element AI, all the more compelling.
Wal-Mart to vendors: Get off Amazon’s cloud (FOX Business): It turns out that large retailers don’t want to fund Amazon’s assault on their businesses anymore, even indirectly. We’ve seen others just choose to run private clouds or go with Google or Microsoft before, but what’s interesting here is that Wal-Mart is trying to make an even bigger dent by forcing suppliers to move. A big question might be whose business they care more about having: Wal-Mart’s or Amazon’s? FWIW, Barb Darrow (Fortune) and I also discuss this in more detail in the latest podcast.

Sponsor: CircleCI
Sponsor: CircleCI
Listen to the new ARCHITECHT Show podcast
In episode 25(!) of the ARCHITECHT Show, Matei Zaharia  and Peter Bailis discuss how the DAWN project at Stanford University is trying bring the power of AI and machine learning to a broader set of users. The project’s initial focus is on making it easier to collect, process and analyze enough high-quality data to take advantage of today’s leading models and algorithms. Aside from the technologies, the Zaharia and Bailis also discuss how they’re working with industry on this project and the right path for moving academic research into the real world. It’s an area Zaharia knows well, having previously helped create Apache Mesos and Apache Spark, and having co-founded Databricks, where he’s still CTO.
Artificial intelligence
Sponsor: Bonsai
Sponsor: Bonsai
Cloud and infrastructure
Sponsor: Cloudera
Sponsor: Cloudera
All things data
Sponsor: DigitalOcean
Sponsor: DigitalOcean
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.