Shelf Life

On Tuesday California legislators passed AB5, a landmark bill that requires companies like Uber and Lyft to treat contract workers as employees. Workers must be designated as employees if a company exerts control over how they perform their tasks or if their work is part of a company’s regular business.

Expect Uber and Lyft’s autonomous vehicle efforts to speed up (as well as their lobbying efforts to gain an exception for its drivers to remain contractors). All those pesky humans, demanding rights like fair wages and health insurance!

On the subject of robots, Simbe Robotics has raised $26 million for autonomous inventory robots to inventory grocery shelves. A VentureBeat story indicates the brick-and-mortar automation market will be worth $18.9B by 2023.

Simbe’s robot, Tally, drives around a space to create a store map and then uses computer vision to “see” what products aren’t on a shelf and identify any missing facings, using RFID for its inventory counts.

VentureBeat reports a single robot can scan 15K to 30K products per hour, compared to the 10K-20K an average human employee can do in 20 to 30 hours. Human, 10-20K, 20-30 hours, Tally, 15K-30K per hour.

Humans will still do the restocking…for now. 

We’ll see how long their shelf life is.

WeIPO

WeWork is finally gonna go public. The company filed today to raise $1B, and reported a $904M net loss on around $1.5B in 1H19. You people gotta start showing up to the WeWorks offices so the WeWorkers can WeWin.

The company has raised over $8B in venture capital since its 2010 founding, so it’s payback time!

For you sports fans out there, you might want to take a second look at Twitter. The company is now testing on Android a way for users to follow interests, as opposed to just other users.

This one is so long overdue it makes my head hurt. However, before you go off RTing, know that the topics will be curated by Twitter and the individual curated Tweets identified by the Russians…err, I mean, by the algos versus editors.

And know that during the time you were reading this post, Nvidia was out breaking more AI land speed records. Specifically, it broke the hour mark in training BERT, one of the world’s most advanced AI language models. Nvidia’s AI platform was able to train the model in just 53 minutes using one of its SuperPOD systems, which consists of 92 Nvidia DGX-2H systems running 1,472 V100 GPUs.

Who said talking to machines is a waste of time!?

Automation Nation

Happy Tuesday.

On the World Cup schedule today: Sweden vs. Switzerland, and Columbia vs. England.

Meanwhile, it’s back to automation nation. Automation Anywhere, a San Jose, California-based robotic process automation startup has announced it raised $250 million in a funding round led by New Enterprise Associates and Goldman Sachs.

According to a report from Venture Beat, Automation Anywhere’s platform employs software robots, or bots, that make business processes self-running:

Using a combination of traditional RPA and cognitive elements such as unstructured data processing and natural language understanding, the system is able to crunch through tasks that normally take hundreds of thousands of human employees.

As to what the company will do with this new round:

The firm plans to spend the bulk of the new capital on deepening customer engagements in North America, India, Europe, Australia, Japan, South Korea, and Singapore, CEO and cofounder Mihir Shukla said in a statement. It also intends to invest in “specialized” machine learning capabilities and new integrations designed to drive “higher operational efficiency” and “flexibility.”

The company’s CEO, Mihir Shukla, was also quoted in the piece as saying that “We believe our Intelligent Digital Workforce Platform can automate up to 80 percent of [an enterprise’s business processes].

That’s far more than its own customers claim can be automated (20 percent).

Austin’s SparkCognition Sparks $24M Series B

Austin Inno is reporting that Austin-based AI firm SparkCognition announced today it has completed a Series B funding round of $24 million.

That round builds on an earlier round last June of $32.5 million. 

SparkCognition builds AI solutions for applications in energy, oil and gas, manufacturing, finance, aerospace, defense, and security. Its website currently lists four core product lines which the company claims provides “human intelligence at machine scale.”

Austin Inno observes that SparkCognition’s new funding is “part of a wave of recent investment in Austin-based AI and machine learning companies,” with large rounds going to CognitiveScale, Conversable, Tethr, and Cerebri AI last year, among others.

Smarter Chips

Couldn’t help but notice these two in-the-same-orbit headlines from Amazon and Google re: their own AI chips.

First, in The Information, it’s being reported that Amazon is developing a chip designed for AI to work on the Echo and other hardware powered by Alexa. 

They report that the chip should allow Alexa-powered devices to respond more quickly to commands, by allowing more data processing to be handled on the device than in the cloud. 

It seems the cloud’s edge is moving back towards the center.

And at Google, according to a post in the Google Cloud Platform blog, the company’s cloud Tensor Processing Units (TPUs) are available in beta to help machine learning experts train and run their ML models more quickly.

Some speeds and feeds deets:

Cloud TPUs are a family of Google-designed hardware accelerators that are optimized to speed up and scale up specific ML workloads programmed with TensorFlow. Built with four custom ASICs, each Cloud TPU packs up to 180 teraflops of floating-point performance and 64 GB of high-bandwidth memory onto a single board. These boards can be used alone or connected together via an ultra-fast, dedicated network to form multi-petaflop ML supercomputers that we call “TPU pods.” We will offer these larger supercomputers on GCP later this year.

That’s a Big Deal!

Happy Monday.

A big deal could hardly wait for Monday morning.  In fact, the biggest tech deal ever.

Broadcom Ltd. has offered roughly $105 billion for Qualcomm Inc., reports Bloomberg, “kicking off an ambitious attempt at the largest technology takeover ever in a deal that would rock the electronics industry.”

According to the report, Broadcom has made an offer of $70 a share in cash and stock for Qualcomm, a 28% premium for the world’s largest maker of mobile phone chips.

If the deal were to go through and be approved, this would make Broadcom the third largest chipmaker, behind Intel and Samsung. And as Bloomberg estimates, the deal would dwarf Dell’s $67 billion acquisition of EMC in 2015.

VentureBeat writes that “the deal would give Broadcom a foothold in the mobile communications market” but comes at a time Qualcomm is still trying to close its own $38 billion bid for NXP Semiconductors, which Qualcomm wants to use to help it get into self-driving technology.

And as VentureBeat points out, Intel is certainly not sitting still as competitive pressure emanates from the likes of Nvidia who are moving into the growing field of machine learning and artificial intelligence, both of which are demanding higher performance semiconductors.

No word at press time as to whether billionaire Prince al-Waleed bin Talal, a prominent member of Saudi Arabia’s royal family and one of the world’s wealthiest men, will be making an investment in this new venture.

The Prince was detained by Saudi authorities on Saturday night, according to the Wall Street Journal, along with at least 10 other princes and more than two dozen current ministers in the Saudi royal family. Mr. al-Waleed is a top investor in tech companies, including Apple and Twitter, and faces charges of money laundering.

 

Home of the Whopper Fail?

“Okay Google, what is the Whopper Burger?”

And that is how the gauntlet was thrown down on the new battlefield yesterday for garnering advertising eyes…errr, ears… in the home assistant device age.

The very same company that encouraged Facebookers to delete their Facebook friends just to get a free burger, and whose mascot who strangly appeared in the corridor with Justin Bieber just before the Manny Pacquiao/Floyd Mayweather fight, has taken guerrilla marketing into the AI age.

First, a little on how Google Home works. Like it’s progenitor, Google Home has a trigger phrase whereby it starts to listen to its owner. In Google’s case, it’s “Okay Google…” followed by the person’s request.

So Burger King figured it would get some free digital media by building some TV ads that made a call out to the Google Home device, whereby it said “Okay Google, what is the Whopper burger?”

To which one would logically ask, from whence came the answer?

In Burger King’s case, reports The Verge, they decided to use the Wikipedia entry, which Burger King apparently edited to read as follows:

“The Whopper is a burger, consisting of a flame-grilled patty made with 100 percent beef with no preservatives or fillers, topped with sliced tomatoes, onions, lettuce, pickles, ketchup, and mayonnaise, served on a sesame-seed bun.”
– via The Verge

Never mind the fact, The Verge observes, that it sounds an awful lot like ad copy, or that just about anybody (Ronald McDonald, anyone?) could go and edit it on a whim.

To make this even more “meta,” the “Whopper Burger” Wikipedia entry now has a reference to this whole escapade:

On April 12, 2017, Burger King released a new commercial, in which an employee states that he had to find a different way to explain a Whopper because they only had 15 seconds, after which he states “OK Google, what is the Whopper burger?”. The dialogue was designed to trigger voice searches on Android devices and Google Home smart speakers configured to automatically respond to the phrase “OK Google”.[81] The specific query causes the device to read out a snippet sourced from Wikipedia’s article on the Whopper. However, prior to the ad’s premiere, the article had been edited by a user who was believed to have ties to the company, so that Google’s automatically-generated response to the query would be a detailed description of the Whopper burger that utilized promotional language. The edits were reverted for violating Wikipedia’s policies discouraging “shameless self-promotion”.[82][83] Furthermore, the snippet became the target of vandalism; at one point, the relevant section listed the sandwich’s ingredients as including “rat meat” and “toenail clippings”, and some users reported that Google Home had relayed information from vandalized revisions.[84][85][81] A few hours later, Google disabled the ability for the ad to trigger automatic voice detection on these devices, preventing the promotional query from being read. Wikipedia also semi-protected the Whopper article to prevent the promotional descriptions or vandalism from being re-inserted.[84]
– via en.wikipedia.org

I kind of gave away the denouement there at the end — Google caught on to the cunning King of the Burger and, before it could spend all that money from all those hard-earned Whoppers on its TV media buy, whose spots would set Google Home assistants a burgerin’ across the country, Google disabled the ability for the ad to trigger the automatic voice detection.

“Okay Burger King, what do you do now???”

I guess they can just bask in the glory of their short-lived PR stunt, which brought far more attention to the Whopper than any Google Home assistant was ever likely to land.

Then again, the ultimate joke may just be on Google. The good Burger King PR for being so clever could very well rub off negatively on the broader home assistant market.

Burger King, I’ll gladly pay you Tuesday for a Google Home assistant today.

And for the record, all this craziness is precisely why I bought the Amazon Tap, the device that we humans have to hit a button to actually turn the thing on.

You know, that old-fashioned idea of the man actually controlling the machine?