AI

Inputs to Outputs

Our lives are increasingly managed by software, which makes it tempting to trust the algorithms and focus on other things, like: why haven’t I had coffee yet? But outputs are only as good as inputs. When bias creeps into training data for machine learning and AI, it can amplify existing inequalities. 

Top of mind 

Facial recognition has recently made headlines, so let’s start there. In a 2019 study, researchers at Brookings estimated that the subjects of common facial recognition training datasets are 75% male and 80% white. Also last year, the National Institute of Standards and Technology (NIST) found “demographic differentials” in performance across the majority of 189 facial recognition algorithms from 99 vendors. 

• For one-to-one tests, such as unlocking a phone or checking a passport, algorithms generally had higher false positive rates for Black and Asian faces. Translation: There’s a higher likelihood another individual could pass as you.
• For one-to-many tests, such as identifying a face from a database of photos, algorithms tended to produce higher false positive rates for Black women. Translation: There’s a greater chance you could be identified as a person of interest for a crime you weren’t involved in. 

As the NIST noted, facial recognition isn’t a monolith: “Different algorithms perform differently.” Though the technology continues to get more accurate, a growing chorus of voices are calling for strict legal guardrails. 

A sampling of other biases in algorithms

• Last October, a study published in Science found “significant racial bias” in a commercial algorithm that guides healthcare treatment decisions for millions of U.S. patients.
• A 2016 ProPublica investigation discovered racial disparities in criminal risk assessment software used by U.S. courtrooms to predict recidivism rates and guide sentencing decisions. 
• Amazon scrapped an internal machine learning recruiting engine after realizing it favored men’s resumes.  

Looking forward

There are reasons to be optimistic. “Fairness,” “bias,” “consent,” and “ethics” have rapidly entered the lexicon of AI researchers. Tech teams are challenged to avoid the path of least resistance by using diverse datasets and auditing algorithms before shipping. There are also calls to diversify technical teams that have remained stubbornly homogeneous. 

Bottom line: AI is a tool with few creators but billions of users. There’s low tolerance for faulty tech with real-world impacts. 

+ For more on the topic, many of you recommended Cathy O'Neil’s Weapons of Math Destruction. Or, for a shorter read, check out Vox’s excellent article on an algorithmic bill of rights. 

HOLIDAY

Who’s Made Juneteenth Official?

Square, Spotify, TikTok, Twitter, Uber, and many more tech companies have publicly committed to observing Juneteenth. Google and Amazon have urged employees to cancel meetings planned for today.  

In other news about tech's response to racial injustice...

Last week, Google banned ad targeting for housing, employment, and credit that’s based on zip code. The company says it will aim to fill at least 30% of its leadership positions with underrepresented groups by 2025 and vowed to hire more Black executives. YouTube has committed $100 million to “amplify” Black creators and artists. 

Yesterday, Facebook COO Sheryl Sandberg said the company will spend $200 million to support Black-owned businesses and organizations.

Netflix CEO Reed Hastings and his wife Patty Quillin announced a $120 million donation to Historically Black Colleges and Universities. “HBCUs have a tremendous record, yet are disadvantaged when it comes to giving,” Hastings and Quillin said in a statement. 

Following two former Black employees speaking out about pay disparities, Pinterest CEO Ben Silbermann told employees he’ll hire outside experts to review employee compensation, Bloomberg reports.

SPONSORED BY ATHLETIC GREENS

Scoop, Mix, Slurp

With Athletic Greens, filling the nutritional gaps in your diet is as easy as that—and we like making things easy around here (like catching up on business news).

The beauty of this delicious greens powder is the amount of whole food-sourced goodness that goes into every scoop.

We’re talking about:

• Vitamins & Minerals. Crucial for a well-functioning nervous system and immune support. 
• Prebiotics & Probiotics. The right stuff to support—and maintain—excellent gut health.
• Adaptogens & Antioxidants. The stress-managing dream team.
• Digestive Enzymes & Mushroom Complex. These little guys support the digestive process and may enhance nutrient absorption.

Top athletes everywhere—including newsletter writers who would’ve gone pro if business news hadn’t gotten in the way—drink Athletic Greens as part of their daily regimen.

Get a free 20-count travel pack with your order today.

Q&A

Connecting Talent and Tech

Per Scholas is a national tech training nonprofit active in 14 cities with 10,000+ successful IT job placements to date, according to the company. I recently caught up with Damien Howard, EVP of Social Ventures, to chat about: 

• Building a diverse talent pipeline with high retention rates (retention of underrepresented groups is a perennial challenge for Big Tech). 
• Making non-adversarial, business-focused pitches to corporate America on the merits of hiring from Per Scholas. 
• Teaching soft skills to reduce culture shock when an employee enters a new industry. 

My favorite quote from the interview: “Our students are problem-solvers due to their history of dealing with adversity. They come with this grit that can’t be taught.”

. And if you’re a tech employer, check out the Per Scholas site. 

BITS & BYTES

Stats: The Internet Association, a trade group repping 41 U.S. tech companies, published its first Diversity & Inclusion Benchmark Report last November: 

• 70% of members have leaders and dedicated budgets for D&I, which correlate with a higher representation of women and minorities in their workforces.
• As internet companies grow, they struggle to retain talent from underrepresented groups (see my convo with Per Scholas’ Howard on how to address that).
• 60% of members publish workforce diversity metrics, which generally show low diversity rates among technical teams and upper ranks. 

Quote: “When people feel motivated and they feel taken care of and they feel like their differences are being celebrated, of course, they'll work harder. Of course, they'll be more excited to think outside of the box."—Morning Brew Head of People Operations Kate Noel on our Business Casual podcast. Check out the full episode. 

Read: Black startup founders face self-doubt and subtle slights in Silicon Valley, Bloomberg reports. 

WHAT ELSE IS BREWING

News that’s unrelated to today’s theme... 

• Audio tweets!
• Zoom will offer end-to-end encryption to all users, 180-ing from an earlier plan to only offer it to paying customers. 
• ByteDance booked nearly $5.6 billion in revenue in Q1, per Reuters.
• Ford is launching an active driver assist feature with camera-based driver attention monitoring. This announcement caught me by surprise. 
• The U.K. is abandoning its attempt to create a contact-tracing app and will use the Apple-Google framework.
• Budget 5G phones are coming, thanks to a new Qualcomm chip. The company also announced a new chipset for 5G-compatible robots and drones. 
• Match Group joined Spotify, Epic Games, and Basecamp in officially denouncing Apple’s App Store fees.
• Huawei is delaying production of its new smartphone, Nikkei reports.

SPONSORED BY QUICK BASE

One thing spreadsheets don’t excel at: people. When you rely on complex, manually-updated spreadsheets, you’re always one keystroke away from disaster. Quick Base changes that. Migrate your spreadsheets to Quick Base and make all your processes people-friendly. With this low-code platform, you can greatly reduce your vulnerability to user-error, while improving data accuracy and security. .

ICYMI

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

• Boston Dynamics Starts Selling Spot Robot & Boston Dynamics Trivia
• A Q&A With Duo Cofounder and Cisco GM Dug Song
• Snap Announces New Tools, Partnerships, and Engagement Data at Developer Conference