DataDownload: Why Facebook’s new privacy features are misleading… at best
DataDownload: Why Facebook’s new privacy features are misleading… at best A weekly summary of all things Media, Data, Emerging Tech View this email in your browser
Today I’m thinking about media and tech, and the unintended consequences of what can happen when you set them loose. Chris Wiggins — an associate professor of applied mathematics at Columbia University and the Chief Data Scientist at The New York Times was on my panel at NewsLab ’20 last week. He’s a fan of history — and so as we think about what AI in Local News can do, he had a sobering reminder of past “Dumpster Fires” and the need for caution and consideration. He gave one example of a Microsoft chatbot that became a racist angry beast in just 12 hours.
The point is — technology will amplify our base nature, and some of our least appealing traits and leaders, unless we build thoughtful guardrails. So — we’re excited both about the tools we’re building, and the people who shape and guide them.
As always — please read, comment, and send along your thoughts about how we can work together. Always excited to hear from you.
The NYC Media Lab Must-Read NewsLab ’20 Gathers Human Brainpower to Ponder Roles for AI in Journalism, Media Industry
NYC Media Lab and Knight Foundation spearheaded NewsLab ’20, which brought together 140 media, academic, and tech leaders to discuss the intersection of AI and local news. Speakers included NYCML’s Steve Rosenbaum, Knight Foundation’s Paul Cheung, and RLab’s Justin Hendrix.
Absent were discussion of “doom and gloom” and “crazy killer robots.” Or AGI for that matter. Meredith Broussard, data journalism professor at NYU, made an important distinction that helped encapsulate the tone of the discussions: it’s “AI to do journalism,” not “AI to make money.”
Localmedia.org presents an excellent summary of the event’s talks, including Jennifer Choi’s nuanced look at the disparity present when news orgs of different sizes collaborate; and John Keefe’s thoughts on avoiding ethical traps that tend to spring up when you reduce the world to “a line in a dataset.”
10 min read
Read More The iPad Turns 10: A Look Back at Its First Decade Wired tracks the evolution of the iPad over the past decade, year by year, from the initial hyperbolic promises of 2010 to today’s mixed bag of results. We love this quote from Steven Levy from around the first launch: “the zeitgeist excitement needle on this gadget has moved past Hula Hoop and Lady Gaga levels, and is approaching zones previously occupied only by the Beatles and the birth-control pill.”
Looking for an even more candid look? Stratechery’s Ben Thompson details how the iPad — meant to fill the gap that smartphones and laptops couldn’t — fell short of revolutionary and became “an inferior Mac.” The core of the argument lies in Apple making it difficult for developers to earn a positive ROI on iPad apps, effectively abdicating access to a sustainable business model.
15 min read Read More For the Media The Scroll Subscription Service Is an Ingenious Web Technology Hack
Too good to be true? Maybe. What if a company decides to acquire Scroll — what happens to the user data? Also, sites like WSJ and NYT are still paywalled. Bohn goes over the pros and cons, and the cool hack the startup uses to serve you ad-less content.
9 min read
Read More If a Novel Was Good, Would You Care If It Was Created by Artificial Intelligence? Will AI write the next passable screenplay — something startup ScriptBook predicts will happen in five years’ time? Regardless if we get the next Jumanji or stitched-together AI Dungeon 2 narratives, it’s a good time for publishers and producers to start considering the future of fiction amidst language-generating algorithms.
“A shift to AI-generated novels could only ever be a short-term strategy. As Barthes intuited and OpenAI’s latest algorithm demonstrates, it’s certainly possible to assemble writing from other writing. But even if this patchwork prose becomes ‘better than human writing’, it would be only drawing on a finite well of inspiration.”
6 min read Read More Facebook Enables Confusing ‘Off-Facebook Activity’ Privacy Tool
In a recent blog (published on Data Privacy Day, no less), Mark Zuckerberg announced new privacy features, particularly an Off-Facebook Activity tool. Zuckerberg said that the tool “lets you see a summary of the apps and websites that send us information about your activity, and clear this information from your account if you want to.”
Gizmodo promptly shovels any misleading language out of the way: “The only thing you’re clearing is a connection Facebook made between its data and the data it gets from third parties, not the data itself.” But there are clauses that confuse even Gizmodo: namely, even if you hit the new “clear history” button, businesses will continue to send data to Facebook.
4 min read
Read More What We’re Watching How Virtual Reality Turns Students Into Scientists
A TED Talk with Jessica Ochoa Hendrix, founder at educational game company Killer Snails. Hendrix demonstrates how low-cost VR helps bring science to life in the classroom, and describes a project that lets students explore underwater ecosystems as if they were marine biologists, along with other careers that they wouldn’t have been exposed to.
6 min watch Watch Now Jobs & Events Job: Manager: Comms and Events
The Communications Manager is responsible for internal and external communication strategy and initiatives for NYC Media Lab’s communications and marketing practice. Apply Here.
Job: Associate Director, 5G EdTech Challenge
The Associate Director will provide leadership and strategic direction for NYC Media Lab in its execution of the Verizon 5G EdTech Challenge program by co-designing and developing a new 5G curriculum. Apply Here.
Event: Data Future Lab’s Next Round program
Deadline: February 10
The Data Future Lab is accepting applications to its 2-year incubator program for Seed to Series A AI startups changing the world and looking to access world-class university resources. Apply Here. A Deeper Look GPT-2 and the Nature of Intelligence
“GPT-2 has no deeper understanding of human relationships than ELIZA did; it just has a larger database. Anything that looks like genuine understanding is an illusion.”
Ready for a bit of philosophizing? Robust.AI CEO Gary Marcus believes GPT-2 flies in the face of everything Noam Chomsky believes about language: there’s no universal grammar (Chomsky and his proponents believe babies are born with a UG embedded in their heads), it doesn’t understand parts of speech, and there’s absolutely no syntactic tree structure to help construct sentences.
Instead, GPT-2 hopes to closely follow John Locke’s view: that learning and experience is all that’s required to develop intelligence. Does it succeed? Through a meticulous and entertaining dive into GPT-2’s capabilities, Marcus shows why the system fails spectacularly, in that massive amounts of money and resources have been spent to produce what amounts to an ELIZA Effect.
17 min read
Read More How AI Is Battling the Coronavirus Outbreak The Wuhan Virus didn’t get announced first by WHO, or the CDC, but Canada-based BlueDot, an AI-powered platform that parses foreign-language news, animal and plant disease networks, and official announcements to warn users in advance.
It spotted signs of the Wuhan outbreak and sent its customers a notification as early as December 31, nearly a week before the CDC’s announcement. Vox takes a look at how AI systems like BlueDot can not only inform epidemiologists and officials when an outbreak occurs, but also predict them.
6 min read Read More Transactions & Announcements Securiti.ai Scores $50M Series B to Modernize Data Governance
Merantix Announces EUR 25M Venture Studio Fund in Davos
Miovision Closes $15M Investment