Asking hard questions of Google’s neural nets

What is the point of tech journalism? Actually, what is the point of journalism writ large in 2015?

Let’s start with the latter – as Matt Bruenig helpfully noted on his blog a while back, many journalists insist that they “work for the public” by covering hard stories like the rationale behind ISIS or, to cite a classic example, the secrecy of the Pentagon Papers. Journalists seem to imagine a wall between themselves and the ‘lesser’ folks who work in fields such as content marketing and especially public relations. Those people might write merely to please other businesses or spin new products, but we journalists – the journalists seem to say – we are fulfilling a higher calling.

Whatever. Jennifer Pan penned a great article for Jacobin about the PR/journalism row, looking at the extraordinary, thankless emotional labor expended by non-“journalists,” mostly women. This work is every bit as “real” as Wesley Lowery going to Ferguson so that he could then endlessly brag about it on Twitter, and it is increasingly the norm for white collar workers everywhere who, overwhelmed with task, are under the gun to channel their emotions into deadline-driven writing.

The plight is shared. Trying to draw these particular lines is just another form of capitalism’s shameful culture of work-prestige, in which we humiliate workers doing thankless jobs to survive, and especially the unemployed who, even if they don’t want or need to work, face unbelievable cultural scorn simply because their efforts are not remunerated! Some day this attitude will be seen for the antisocial embarrassment that it is.

In this context, “journalist” becomes a moral and normative, rather than a descriptive, term.”Journalist” is to the language-intensive fields what “entrepreneur” is to the tech sector: a loaded word for dressing up ultimately mundane labor, which is everywhere. The entrepreneur runs businesses, and the journalist writes words that target a specific demographic favorable to advertisers. Doesn’t sound so extraordinary, does it?

Are Vox, BuzzFeed, and even The New York Times now anything more than vehicles for precisely paired ads and text? Look, here’s a mansplainer about coal power plants, alongside a tasteful banner ad for Hulu. Journalists imagine the traditional wall between their work and the business itself, but even this separation has its limits. I doubt The Verge is going to do a deep-dive series on the merits of Leninism or on the unremarkable hypercapitalism of Google, Facebook et al.

Which brings me back to the original question – what do tech journalists do?

The likes of The New York Times’ Farhad Manjoo or even one-man operations like Stratechery’s Ben Thompson and others from the Web’s most trafficked sites often create work that is indistinguishable from press releases. I mean, Manjoo once talked about how Spotify encourages “a musical culture that more closely resembles a cocktail buffet than a sit-down dinner,” without asking if this is a good thing, why it happened, or what is means for the people producing the music. Thompson’s posts, while typically more nuanced, rarely challenge what firms from Uber to Apple do, instead viewing these companies as essentially apolitical entities that move with all the impersonal intractability of something like the weather or time itself. A typical paragraph reads:

“When you think about it that way — that Netflix isn’t so much a network as they are a type of marketplace in which consumers can give their attention to creators — it becomes apparent that Netflix isn’t that far off from Uber or Airbnb or any of the other market-makers that are transforming industry-after-industry.”

Netflix isn’t just some company that put videos on its servers, no – it’s one of the new “market-makers,” an enviable godlike position in a world in which “the market” is still treated as a fact of nature rather than, as John Pat Leary has so elegantly noted, “a political instrument and a historical notion.” Companies are really just people – laborers, beholden to capitalists – but tech journalism makes them seem like persons – i.e., autonomous, cohesive entities with their own histories and personalities. The U.S. Supreme Court would be proud.

To get a real sense of just how completely tech journalists and reporters have capitulated to the propaganda, though, you have to have read its recent hysteria about Google’s neural nets. Basically, Google fed random images to its custom cloud infrastructure and got some “weird” results that the tech press thought were cool. Here’s Business Insider’s hot take:

“Google’s artificial neural networks are supposed to be used for image recognition of specific objects like cars or dogs. Recently, Google’s engineers turned the networks upside down and fed them random images and static in a process they called “inceptionism.”

In return, they discovered their algorithms can turn almost anything into trippy images of knights with dog heads and pig-snails.”

Bleh. Even the grouchy Gizmodo couldn’t muster much more than some linked tweets and images and an invitation to try it out for yourself. So Freddie de Boer was right when he noted:

“I still have not seen a single piece on this “Google’s dreams” thing that asks a single hard question or brings an ounce of skepticism.”

So in that spirit, I’m going to ask some hard questions of my own (I figure my skepticism is more than implied throughout this entire piece):

  • What does this mean for facial recognition technology? Is Google preparing its neural nets to catalog all human faces with identities? What if someone doesn’t want this?
  • Google is an advertising company. Is it planning to run psychology experiments on its users by showing them subtly “modified” images that are just click-traps?
  • We hear a lot about how Google, Microsoft, and others face a “skills gap” that prevents them from hiring U.S. workers. This is bullshit, but how much do these companies actually benefit from using this vulnerable labor from India and other places to fuel the development of questionable, antisocial projects, like these neural nets?
  • We’re often told that superhuman AI is not only inevitable, but necessary. If these ‘dreams’ are the basis for what we’re creating – super fast image recognition – is there really so much promise in this technology, though? Better ads, quicker load times – that’s it?

I could keep going, but all of this is tedious I realize. Just as journalists are increasingly at the mercy of advertisers – to the point that even great outfits like iMore are inseparable from the annoying ad engines that keep their lights on – tech journalists are more subtly ruled over by the motives of reified firms that seemingly none of them dare question.

%d bloggers like this: