I am taking a little time off to rest and recover after publishing well over 100 issues and nearly 40 podcasts this year, so this is the last issue of the newsletter until Aug. 14.
For that reason, and because it was a slow news day, I’m going to leave you with a collection of the most-popular ARCHITECHT content so far this year. A lot of new subscribers or listeners might have missed these pieces, and even loyal readers/listeners might have missed a few things spread across this newsletter, the podcast and the homepage (where I reprint newsletters, publish podcast highlights and occasionally even write original content). And as a reminder, here’s a list of links for subscribing to the podcast pretty much anywhere.
The startup co-founded by one of deep learning’s godfathers announced $102 million in funding last month, and interest has been off the charts:
“In the month after the funding announcement, Element AI received 150 qualified applications to join its network of partner startups, 200 inquiries from potential customers, and 1,500 resumés. During that time, the company created roughly 50 sales proposals for how it could help those customer applicants.
"CEO Jean-François Gagné recently finished a trip to France, where he met with 10 of the country’s 40 biggest companies. All of those meetings led to sales proposals.”
There’s some sound advice in here, which mirrors what I’ve heard recently from experts (including Jeremy Howard of Fast.AI). Basically, the core models and approaches are available, solid and learnable, so find people who know your data and can work around the edges of AI without being experts.
It sounds like a lot of jargon, and it is, but transfer learning will be critical to the broad applicability and adoption of AI. If a system can’t learn to deal with new domains, or even changes to its existing space, it’s not very useful at all.
That’s a 9 percent increase over last year (which was only a 3 percent increase over the year before), but data center sales are actually Intel’s slowest-growing area (ignoring the one business that’s losing revenue). When you compare Intel to AWS, it starts to become pretty clear how the power dynamics of data center hardware are evolving.
Or so say people familiar with the matter. Accel, which has been an investor since it led the company’s seed round, and SoftBank led the round. Aside from its incredible popularity as a collaboration tool, I think Slack also could become a very important tool in automating data center operations and certain software development tasks. There are always lots of rumors about Slack getting acquired, and with a new $5 billion valuation, the bar is set pretty high.
This piece reminds me of the advice that Facebook CSO Alex Stamos gave in his Black Hat keynote, which is that there are basic things still broken in many organizations that would prevent a lot of damage if fixed. Also, this medical analogy on simplicity is good (and informs a lot of the discussions I have with peopled doing applied AI and ML, too):
“A month or two ago I was having a discussion with a physician about obscure diseases—commonly referred to as zebras. While I was considering these zebras in the context of effective data mining strategies for medical diagnosis, he made an interesting point. One of the things that they teach new physicians is the phrase ‘When you hear hoofs, think horse, not zebra.’ The principle is quite simple—the odds are the patient has the more common diagnosis than a rare, improbable one.”
ARCHITECHT delivers the most interesting news and information about the business impacts of cloud computing, artificial intelligence, and other trends reshaping enterprise IT. Curated by Derrick Harris.