DataDownload: Amazon is back in NYC in a big way! China bans all deepfakes…

NYC Media Lab
6 min readDec 7, 2019


DataDownload: Amazon is back in NYC in a big way! China bans all deepfakes… A weekly summary of all things Media, Data, Emerging Tech

Ok, raise your hand if you’ve been to at least four holiday parties in the past week? Just a note — pace yourself ok? For me, our Machines + Media Working Group drinks were pretty fantastic, but the ODP Holiday party was off the hook. Next week… expecting holiday shindigs to be wonderfully exhausting.

Meanwhile — the history of Neural Networks according to Ars Technica. China outlaws Deepfakes. Sacha Baron Cohen takes on Sergey and Larry. And Amazon — a millisecond after bailing on NYC — is back in a big way. Not Long Island City, but Hudson Yards. Because, of course, they have to be in NY — the tech talent and University resources are just too fantastic to ignore. Always good to see NY’s charms make its alluring case.

Public service announcement: please don’t drink and drive.

As always, ping me if you have thoughts or feedback about the Newsletter.

Steven Rosenbaum
Managing Director
NYC Media Lab


How Neural Networks Work — and Why They’ve Become a Big Business

Who doesn’t love an entertaining history of neural networks — especially one done by Ars Technica and complete with a parable of a fictional software company who’s app rates objects as “hot dog” or “not a hot dog”? From Perceptrons to AlexNet to Tesla, Ars covers a timeline from the 50s to tomorrow.

We found the most interesting bit to be the justification behind AlexNet’s 2012 breakthrough — namely, why an AI revolution wasn’t sparked earlier, despite GPUs, image scraping, and underlying algorithms all being available for years… “apparently no one had thought to combine them.”

Facebook’s Head of AI Says the Field Will Soon ‘Hit the Wall’ Wired senior writer Will Knight met with Facebook’s VP of AI Jerome Pesenti for a (somewhat stiff) conversation on disinformation, synthetic media, and AGI. Pesenti and Knight also discussed the inevitable compute wall (which Pesenti believes we’ve already hit in certain areas), and the future of Facebook’s AI products.

On the products end, Facebook previously shuttered its M chatbot. Pesenti didn’t divulge specifics on future releases, just that they’re working to create a safer, more personalized platform.

For the Media

These Reporters Rely on Public Data, Rather Than Secret Sources

“You can be on your couch in front of your computer and solve a mystery of a missile system downing a plane.” — Aliaume Leroy

Previous generations have romanticized the investigative journalist’s physical detective journey. But with the proliferation of public datasets and a deluge of social media clues hiding untold stories, “following the money” has morphed into “following the data.” Open-source journalists now work on laptops to uncover the world’s most pressing mysteries.

Eliot Higgins is one of the most prolific members of the movement, founding open-source news side Bellingcat in 2014 (fun fact: he was inspired by — and attributes his success to — video games, in that he figured any mystery can be solved with enough prodding). Higgins and fellow open-source journalists rely on a cadre of tools, from Google Earth to image verification; trust doesn’t come from brand but transparent methodology, with journalists walking through their process publicly (see Aliaume Leroy’s excellent thread below).

Cards Against Humanity Writers Are Battling an AI to Keep Their Jobs, and You Can Watch

For Black Friday last week, Card Against Humanity writers challenged an AI to come up with the better-selling card pack. Holed up in a meeting room, the writers came up with their lurid material while an old-school printer clunked out its own travesties (which we can’t even reprint here).

Surprisingly, CAH actually used a neural network (GPT-2) instead of pretending to train an AI, and fed 76k cards into the pretrained network to come up with its own salacious writer.

China Makes It a Criminal Offense to Publish Deepfakes or Fake News Without Disclosure

A new policy in China banning the use of false information or deepfakes without proper disclosure will go into effect on January 1, 2020. While California was the first state to criminalize the nefarious use of synthetic media in political cases, the law was aimed at mitigating potential damage in elections and excluded news and entertainment. China, on the other hand, has a broader reach — targeting any creator and hosting service violating the rules.

What We’re Watching

ADL International Leadership Award Presented to Sacha Baron Cohen at Never Is Now 2019

Many consider the former Ali G star of mankini-wearing fame an unlikely revolutionary in the fight to temper the power of big tech. But it’s worth noting that Sacha Baron Cohen studied history at Cambridge University and wrote a thesis on the American Civil Rights Movement.

His speech contextualizes the current dangers of unregulated platforms around civil rights, prejudice, and racial violence, and it is a powerful one. If you’re having trouble finding time for the full 24 minute version, you can also check out this three minute recap (but seriously, it’s worth watching the entire thing!)

Events & Announcements

Event: Disinfo 2020 — Prepping the Press
Date: December 10, 9AM-3PM
A day-long conference from Columbia Journalism School exploring how disinformation will affect the 2020 election — and what we can do about it. Register Here.

Event: Fireside Chat w/ Director at Tableau Software
Date: December 12, 6PM-8PM
A Fireside Chat with Alan Stein, Director of Worldwide Technical Support at Tableau Software. Alan leverages his analytical proficiency, honed on Wall Street, to quickly identify critical business levers, develop operational strategy, partner cross-functionally, and execute on plan to consistently exceed goals. Register Here.

Event: Smart Communities — Smart Futures w/ Tech Innovation
Date: December 16, 6:15PM-8:45PM
Smart cities stay safe and operate efficiently. Discover the smart technologies that keep your community connected. Register Here.

A Deeper Look

China Uses DNA to Map Faces, With Help From the West

Chinese scientists are using DNA samples obtained from ethnic Uighurs (along with millions of samples from Xinjiang) to attempt to recreate images of faces for potential souped-up surveillance systems. Meanwhile, the scientists working on the technology are receiving funding from European institutions and getting their work published in respected journals.

The recreation process, called DNA phenotyping, has been around for years, but the practice isn’t perfect: “it often produces facial images that are too smooth or indistinct to look like the face being replicated.” The NY Times investigative piece profiles Tang Kun, a specialist in human genetic diversity at the Shanghai-based Partner Institute, and his involvement with Chinese police on the phenotyping program.

Amazon Leases Midtown Space after Halting HQ2 Plan for NYC

When Amazon abandoned plans to build HQ2 in New York’s Long Island City, critics and activists were surprised and delighted (and there were many). But this didn’t mean the tech giant fled the city. The company recently signed a lease for 335,000 square feet of office space in the Hudson Yards neighborhood, set to open in 2021 and accommodate 1,500 workers.

Transactions & Announcements

WPP, Bain Complete Kantar Transaction

Alphabet Pushes Duolingo To $1.5B Valuation

Intel In Talks To Buy AI Chip Startup Habana Labs



NYC Media Lab

NYC Media Lab connects university researchers and NYC’s media tech companies to create a new community of digital media & tech innovators in New York City.