Dave Gershgorn

Journalist

Dave Gershgorn is a writer and photographer based in New York City. He serves as a reporter for Quartz, with a personal focus on translating artificial intelligence research to a mainstream audience. He was previously the Assistant Technology Editor of Popular Science. When Dave is a photographer, he likes to shoot long, character-driven features. His work has been featured in QuartzPopular ScienceThe New York Times and the Wall Street Journal.

 

On the weekends, Dave likes hot dogs and any other kind of dog. 

By Grabthar's Hammer, what a Monday

Big Monday for AI news:

OpenAI's Universe brings reinforcement learning to any desktop software, which could be used to control DeepMind's new 3D learning environment. But don't forget Facebook, who just open-sourced their library that connects StarCraft to the open-source Torch framework. (Here's a Reddit comment with all the code and documentation.)

Udacity's Oliver Cameron (who is a great Twitter follow) also created a really interesting list of open-source machine learning datasets. One that jumped out at me: an open-source dataset for every military dispute in the last 200 years. Also, mushrooms! It's a fun doc to poke around.

Uber's new AI research lab is well-positioned to focus on squeezing water from the metaphorical stone, and Microsoft is trying to encourage more women in machine learning research.

One issue sometimes cited for the dearth of women in computing fields is the lack of professional role models who could inspire girls to pursue their STEM dreams. We’ve attempted to counteract this by asking 17 women within Microsoft’s global research organization their views on what’s likely to occur in their fields in 2017.
— Microsoft

Japan's parliament is reportedly exploring the use of AI in anticipating and answering questions that might arise in the process of policymaking. The AI would be trained on the last five years of parliamentary data. Good luck with that.

Finally for today, OpenAI's Ian Goodfellow was publicly accosted by noted ghost-of-AI-researcher-future Juergen Schmidhuber, who stood up in the middle of a talk and asked a series of pointed questions about the work. This is seen as pretty bad form, and Schmidhuber seems to be the king of it.  [Don't do it.]

There was just a lot happening today, so I thought some people might enjoy a one-stop shop for AI news. Let me know if you like it, and I'll do it more often!

-Dave

The Chinese phrenology paper is an important reminder of AI ethics and not much more

If you're reading this, odds are you're familiar with the paper "Automated Inference on Criminality using Face Images" from Shanghai Jiao Tong University. If you're not familiar, it's basically some research that uses machine learning as modern-day calipers, with the conclusion that we can judge criminality from facial features.

Obviously, this is not true. It's not even that interesting of a paper, to be honest.

But there are a few things worth mentioning. Foremost, this paper did not go unnoticed. Every day, dozens if not hundreds of ML researchers scour the research repository ArXiv looking for the latest published papers, and people love to call out weird or flawed research on Twitter.

Google engineer François Chollet kind of likens it to crowdsourced peer review.

Great, we caught it. Now what?

Here's the reason I'm writing this on a personal blog and not Qz.com, where I'd usually post (and might put something up Monday). 

This is a good case study on how A.I. research can be unethical, but this just a symptom of having a litany of open, flexible tools. No matter how troubling Sam Biddle finds this work, it was done by two academics and posted on a non-peer-reviewed, public repository. It has not been endorsed by anyone but the authors, and implicitly by the university which let them publish under their name. (To me, that's the most troubling aspect of this whole ordeal.) Major players in field of artificial intelligence have spent considerable resources to ensure machine learning is easy to access.

I understand artificial intelligence can seem scary and intimidating, but this AI is no more to blame for this than calipers and equations used in phrenology. 

Biddle's story does a good job of summarizing why phrenology is bad and by extension, why this paper is bad. He's right! But at the same time, the story paints the industry in broad strokes.

The story begins:

The fields of artificial intelligence and machine learning are moving so quickly that any notion of ethics is lagging decades behind, or left to works of science fiction.

Incidentally, today New York University hosted dozens, if not hundreds, of machine learning and artificial intelligence researchers to talk about how to make machine learning more ethical. The panels ranged from raising questions about algorithmic policing to implicit bias in text-based AI. The title of the conference is Fairness, Accountability, and Transparency in Machine Learning, and was sponsored by Google, Microsoft, and the Data Transparency Lab.

Microsoft Research's Kate Crawford, who was quoted in Biddle's piece, said at the FATML conference, "We have a responsibility to say what work is ethical and what work is not."

However, it should be said that this paper isn't without a lesson. Machine learning is an incredibly powerful tool, able to rip through data and draw biased, misleading or just wrong conclusions at a dizzying pace if built incorrectly. Academics and AI researchers reacted exactly as they should. But I don't buy the argument that this research creates some kind of market for this technology that did not exist previously.

The way this story is told is important. The way that people understand AI is important. The way people understand the engineers building this AI is important. All I ask is that we're smart in the way we cover papers like this.

Think I'm wrong? Did I miss something here? Let me know, I'm on Twitter.

The Windows Conundrum

There are a few moments in my life where I've had the wisdom or clarity of mind to take a step back from a task and really try to understand what it means to succeed in it. One of those times was Monday. I was handed a computer with Windows 10 and asked to review it for Popular Science. I'm still thinking about it, but spoiler alert, I wrote the article and you can read it here. I'm writing this because I think it's important to have a public record of my mindset in this case, and if someone takes the time to come to my blog in this odd corner of the internet from that article, they deserve to know what I was thinking.

I use a MacBook Pro most days. In a stroke of irony, I write this from my Microsoft Surface, and wrote my review from my Mac. While I'm familiar with the Windows operating system (in the article I list my Windows chops, including building a computer with Windows 8 and crashing Windows 95 by deleting System32) but I don't use it every day of my life. Sure, I use it most days, the aforementioned Windows 8 computer is my gaming rig, and I love my Surface. But I wondered if I would be doing a disservice to readers by not giving them a review from someone in the trenches of Windows 8. I'm really good at Mac OS X. I know so many shortcuts, and I'm quite zippy. I'm proud of that. I don't feel the same way about Windows. Maybe it was a Windows 8 thing. I have a theory, that has no real evidence behind it besides testimonials that I've read, that Windows 8 was so widely panned because there was no sense of permanence-- no feeling that you were at home or comfortable. It's because you were forced to hover between two screens (home and Metro), and I never felt like the controls were consistent.

That's the biggest fix with Windows 10 that I could see. You feel like you're home. I can't really describe the feeling I got when I booted Windows 10, popped open the Start screen, and scrolled through the application bar. It was cathartic in a way I didn't think software could make me feel. I felt like I had been unease for years, but now things were back to normal.

I had a lot of questions to ask myself, writing a fairly visible piece for potentially hundreds of thousands of readers who wanted to know if their computer would be better served with an free operating system upgrade. Could someone who was a Mac user properly review a Windows OS? Would my personal bias enhance the clarity my observations, because wonky features would stand to be more noticeable coming from a very refined OS? Or my perception of what an operating system should be (because my contentment with OS X) be skewed? And on a third hand, I had no choice.

That feeling I had when I opened the application bar, as silly as that sounds, validated the fact that I could be impartial when testing the system. While I am more comfortable in OS X, I do have a deep understanding and history with Windows. I think success in this case was being able to put out a piece of text, that as someone with a reasonable amount of self-awareness I could send to my editors and not feel a sense of unease. Honesty, in other words. I think I reached that goal. A review is only a review, my opinion. I read about Windows every day for months in preparation, and I like to think my opinion is relatively valid. I liked some things, and truly disliked others (Cortana). I think it's important that people have access to the basis of my opinions, so I wrote this little personal post. If you have questions, I implore you to reach out. I'm always looking for discussion on your experiences.

New site, new focus

You could also call this post "Towing the line: a story of indecision and insomnia." This site now reflects that I've been writing professionally more consistently, which was lacking in its previous form. I've swung back and forth from writing to photographing to writing et cetera, and now there's a healthy medium represented publicly.

 

I'd like to update this blog more regularly, as well, with things I've been tinkering with, background on stories, etc. There's a blog I've been wanting to start called "Dave Tinkers," primarily focused on my projects in software and hardware tinkering. But, I've had some hosting issues, so I've decided to merge it here. It makes sense, because as my writing for work merges with my personal interests in making things, I think it behooves me to show that I'm not just talking the talk.

I might even rant about journalism. Or post a doodle. Who knows? What an exciting possibility.

On the Line at the Westminster Kennel Club Dog Show

In the moment before a dog and handler compete, there is often this moment of open intimacy. In agility trials, some play games incentivized by bits of hot dogs and other treats—in more formal breed judging, moments can be a bit more serene. These images capture a sliver of the deep connection between dogs and those who dedicate their lives to show them.

Published on WNYC: http://www.wnyc.org/story/westminster-dog-show-on-the-line/

Ferguson Protest in NYC: Tuesday

NYC Protest of 'No Indictment' in Ferguson

Journalists/ Coverage in Ferguson

Here are some great sources of information in Ferguson, Mo.:

I suggest the Livestreams, they're really well done and give great perspective.

Vice's Tim Pool Livestream: http://new.livestream.com/timcast/events/3295551

KARG Argus Radio Livestream: http://new.livestream.com/accounts/9035483/events/3271930

NYTimes' Developing Photo Story: http://www.nytimes.com/interactive/2014/08/17/us/ferguson-photos.html

Abe Van Dyke in Ferguson: https://www.facebook.com/vdcphotonews

Twitter List of Journalists: https://twitter.com/Dave_Roadsters/lists/ferguson-news

These are what I'll be following!