GEOPOLITICS

Oh, EU Got What AI Need

Europe’s powers that be have an issue. To simplify it greatly, the continent runs on technology built elsewhere. Brussels wants to change that even more than it wanted to be done with Brexit. 

“Not all those who wander are lost”

This J.R.R. Tolkien quote concerns fellowships and rings, but we’re using it in service of the “Digital Compass.” That's the name of a provision in the EU's $2 trillion pandemic recovery package, which earmarks $150 billion to help the EU find its technological way. Europe’s plan has four key planks:  

1. Train a more technical workforce. Self-explanatory. 
2. Build “secure and substantial” domestic digital infrastructure. Think cloud and AI. 
3. Drive “digital transformation” of businesses. This is government/Fortune 500-speak for ditching dinosaur databases and upgrading tech stacks. 
4. Digitize government. In this department, Estonia is goalz. 

Where’s the compass pointed? The EU wants more digital sovereignty and less dependency on others for key technologies. An example of dependence: As of last March, every top EU bank was primarily using American cloud services. 

So what does progress look like? 

The EU aims to double its global market share of cutting-edge chip production, with the goal of 20% by 2030. One ace the EU does have in its pocket is Dutch semi equipment supplier ASML, which we’ll tell you more about soon.

Semiconductors provide a good window into continental pushes for domestic talent, industry, and infrastructure.

Opening that window: It’s not just the EU sweating about semis. China is spending tens of billions to establish chip self-sufficiency. Biden recently ordered a review of the US’ critical supply chains. As a result, you’ll likely see Washington pour billions into subsidizing state-of-the-art US chip fabs (from TSMC or Intel). 

Zoom out: Believe it or not, Uncle Sam did help sow the seeds of today’s Silicon Valley in the twentieth century. That doesn’t mean top-down industrial policy alone can magically produce a dynamic innovation ecosystem in the EU. It's a start, but the recipe is complicated.

AI

Not Into Labels

Gig workers who label data to train AI systems said they felt pressured to align their answers with the majority view—or risk losing their jobs, according to a Motherboard investigation. 

These “ghost workers” typically earn less than minimum wage to label images and other data on microwork platforms like Amazon’s Mechanical Turk or similar ones run by IBM, Clickworker, Appen, and Lionbridge. 

• “There were many images that were just formless blobs or of basic objects,” said one dataset labeler. “It was quite a stretch to come up with emotions and explanations at times.” 

Back up: Data labeling = big business in the tech world. Companies like Google, Expensify, Pinterest, and many more have used these microwork platforms. 

We’ve written time and time again about bias in data sets, and how AI trained on that data can perpetuate bias with immeasurable consequences. 

Here’s the kicker: Every data set out there is already biased to some degree, as it’s virtually impossible to root out the systemic issues they ultimately reflect. So imagine the heightened level of bias that could be prevalent in a dataset created by workers who are incentivized to prioritize conformity over accuracy. 

SPONSORED BY DOCSEND

The Deck That Brought in the Dollars

This is a story about the perfect pitch deck. It’s how a founder got the ~capital~ he needed to turn his pipe dream into a pipeline for VC capital funding with the help of DocSend. 

It’s the tale of how Alex Simon and Frankie Scerbo founded Elude—the travel search engine. But before his business could take flight, he needed to secure funding, which is where DocSend saved the day. 

DocSend’s insights allowed Alex to track individual investor interest in their pitch deck without wasting time on investors who didn’t even bother to open his deck. His team met with 50 out of the roughly 450 investors contacted, (outperforming the average successful pre-seed fundraise).

They knocked out those 50 meetings in 2 weeks, resulting in Mucker Capital and Unicorn Ventures co-leading their pre-seed round with an initial $650K investment—with the round totaling $1.5M.

Moral of the story: it takes more than a million-dollar idea to secure VC funding. It takes the help of DocSend.

Get the full story here. 

EXPLORATIONS

Inside Amex’s Confidential Fraud AI

Ever gotten a fraud alert from your bank? Behind that simple text or email is a machine learning model based on billions of data points, millions of lines of code, and thousands of hours of human monitoring.

So, for the second piece in our Demystifying Algorithms series, we’re giving you an inside look at Amex's brand-new fraud risk AI—and what it took to overhaul it once Covid hit. 

The specs: Amex’s “Gen X” model, which rolled out to every cardholder in January, runs on a sequence of more than 1,000 decision trees and sweeps more than $1.2 trillion in annual spending. A team of about 30 monitor the model 24/7 and update it at least once a year. 

The snags: The new model’s build and rollout stretched on for nearly a year due to Covid-19—the process usually takes about three months. At one point, Amex even struggled to get the AWS server space it needed to test the model. It was a wild ride full of bizarre pandemic spending patterns, duct-tape fixes, and work-filled weekends. 

Read the full Explorations piece here. 

BITS & BYTES

Stat: 800+ employees are working on Vesta, Amazon’s secretive home robot project, Business Insider reports. The picture above = a different type of home robot Amazon will soon sell. 

Quote: “The more that we can teleport around, not only are we personally eliminating commutes and stuff that’s kind of a drag for us individually, but I think that’s better for society and for the planet overall, too.”—Zuck, riffing on augmented reality, to The Information. 

Read: Results from the Apple Women’s Health Study. 

SPONSORED BY ZENDESK FOR STARTUPS

Unicorn-levels of customer support for your startup. You’re in the midst of growing your burgeoning biz—the important thing now is to not leave your customers hangin’. Zendesk’s latest benchmark study found that Unicorns are 37% more likely to offer omnichannel customer service and 61% faster at adding self-service support in their first year. Zendesk for Startups is your ticket to fewer support tickets. Join today and get six months free.

WHAT ELSE IS BREWING

• President Biden plans to appoint Lina Khan, an antitrust scholar, as an FTC commissioner, Politico reports. The White House also recently hired Tim Wu, a prominent Big Tech critic, to work on competition policy. 
• Dutch and Belgian law enforcement officials have taken down Sky ECC, an encrypted messaging platform popular with criminals. 
• Moneygram terminated its partnership with Ripple, the company behind the XRP cryptocurrency that is currently in a legal brouhaha with the SEC.
• Coinbase shares recently traded in secondary markets at an implied valuation of ~$90 billion, Bloomberg reports.
• Qualcomm’s next CEO Cristiano Amon tells CNET he is quite worried about the semiconductor shortage. 

TRIVIA

At long last, Roblox is going public. This quiz tests your knowledge of the gaming and social media company. Even if you’re not up to snuff on Roblox, this will still be of interest. Trust us, you’ll be hearing the company’s name a lot more going forward…

Take the quiz here. 

TECH THROWBACK

Headlines this time last year were, uh, bleak, to put it lightly. So we journeyed to the WSJ archives of March 2011 in search of something more nostalgic. The headlines don’t disappoint.  

• Netflix and the iPhone would like a word: “NCR to Install Blockbuster Express Kiosks in RaceTrac Stores” / “BlackBerry Maker Mulls More Investments” 
• Wait until they learn about bitcoin: “Safe-Haven Currencies Get Boost From Japan Crisis”
• When you’re an EV before your time: “Nissan Faces Blow to Leaf”

ICYMI

Catch up on the top Emerging Tech Brew stories from the past few editions: 

• Inside Arrival’s summer of EV testing ahead of its UPS deal 
• Report: How AI research, investment, and deployment grew in 2020
• Microsoft’s new central nervous system for mixed-reality apps