Discover more from How Things Work
Life as Grist For the AI Mill
One reporter confronts the algorithmic maw
How Things Work has written about the challenges that AI poses to the media industry and society at large. Today I’m bringing you a guest post on this topic by Dave Infante, a good reporter and union man. A version of this piece first appeared on Fingers, Dave’s independent newsletter about drinking in America. If you’re interested in original coverage from the left on the business, culture, and chaos of alcohol, consider buying a subscription to Fingers — How Things Work readers get 20% off their first year.
Once, I interviewed a craft brewer who’d recently realized he’d been breaking his body in unsafe conditions for terrible pay just for a chance to produce something he loved in an industry that didn’t love him back. “There are people making money here,” he told me, “but it’s not us.”
As an independent journalist whose entire career has played out against the volatile boom-and-bust cycles of the venture-capital-financed, algorithmically fucked media industry, it was a familiar sentiment. This stupid business’s ever-tightening death spiral has taken down hundreds of publications and tens of thousands of writers and editors in the past decade, but there are people making money here. With their founder-brained buzzwords, obsessions with growth, and barely concealed contempt for both editorial labor and its fruits, they’ve never been particularly hard to spot. Now, with the artificial-intelligence land rush tantalizing the investor class with the mirage of a fat-margined, endlessly scalable, labor-free media future just over the horizon, it’s easier than ever.
Which brings us to Volv.
Volv? you may be wondering to yourself. Fair question, dear reader. Let me clear up your confusion: Volv. Does that help? No? Ah. According to its website:
Volv is “the anti-social media app.”
Volv’s “AI curates interesting content across cyberspace and delivers it to you in 9-second articles.”
Volv says “the future of media is here.”
More to the point, Volv is a platform that uses AI software and human editors (more on them in a minute) to summarize real stories written by real people in brief “bits” that users can swipe through like a social-media feed. “Think TikTok for readers and writers,” encourages the company’s site, a pastiche of too-online vernacular and Gen Z cultural references. So unfortunately, we must.
Normally, I simply ignore the schlock that passes for “innovation” in this business, and I would have happily lived out the rest of my days believing “Volv” was just Volvo short a vowel. But the other day, the company tagged me in a few of its social media posts, and I got curious. Here’s the tweet I was included in (it’s since been taken down):
Clunky AI-generated syntax and diction aside, this was obviously an aggregation of a newsletter I’d recently published about some legislative developments on my actual beat. Aggregation is just a part of publishing on the internet, and it can be of real value to the original producer, the aggregator, and the aggregator’s audience when done well. But in trying to jam a pair of complex topics into a blurb that takes just nine seconds to read, Volv’s AI-abetted rewrite made my coverage worse in subtle ways. The blurb—a “bit,” in the company’s vernacular—glazes vital context about the legislation in question, misattributes and obscures the political motives of the (mostly Republican) politicians wielding power on the matter, and ties it all up with the rhetorical equivalent of a shrug. Because of its framing, I also suspect Volv somehow scraped the paywalled portion of my newsletter, though I have no way to prove it.
After tweeting about this little indignity, and DMing Volv’s Instagram account to demand my logo be removed from its app and social posts, I eventually got the attention of cofounder Priyanka Vazirani. In a video interview the next day, she repeatedly denied that Volv indexes paywalled content in any way. “That's probably why you think that the Volv snippet was a bad reflection of your newsletter, because we didn't go through the entire newsletter,” she told Fingers.
How Things Work is a reader-supported publication. If you like it, consider becoming a free or paid subscriber.
Which, yeah, could be! Or it could be the fact that Volv uses an automated ChatGPT-based software to gin up the blurbs, and, according to Vazirani employs just three staff “editors” to oversee all the content the platform publishes. She declined to tell me how much they were compensated, but given that the company has raised “a little over” the $150,000 it received for participating in Snap’s start-up accelerator program, Yellow, in 2021, and Vazirani says her and her cofounder, Shannon Almeida, are not currently drawing income from Volv, back-of-the-napkin, tax-agnostic math works out to… what? $25,000 per editor, per year? “Everybody has different part-time, full-time, intern” job requirements, she responded when I pointed out that that’s hardly any money for a staffer, especially without benefits. “And at the same time, they’re not all in the U.S.”
Volv is hardly the only business that hopes to use AI to produce passably human writing. In the past year, start-ups and legacy firms alike have flocked to the technology, despite well-documented problems with its biases, legality, energy waste, and the conditions of the foreign human labor required to actually keep it running. I’m just writing about Volv here because it barged into my mentions, and represents—to me at least, though certainly not to Vazirani—a common denominator of this new method of top-down digital arbitrage on the value of journalistic labor.
For a story like mine, which is relatively low-stakes in the grand scheme of things, Volv’s inaccuracies are vexing, but not catastrophic. (The fact that they appeared under my copyrighted Fingers logo, falsely suggesting my in-Volv-ment, is a more troubling transgression. The company has removed both the “bit”—their term for article stubs—and all the social media posts about it following my request.) But you can imagine a broader set of consequences stemming from Volv-style editorial inaccuracies and omissions on more pivotal coverage. The news is hard and complicated, and human reporters (very much including me!) regularly make mistakes. But unlike a generative AI model trained on an unknown corpus of copyrighted materials, human reporters are capable of being transparent about biases, sources, and errors. There’s an entire professional ethics code that demands it, in fact!
A tried-and-true way to publish quality reporting—or even just “content” unfettered by the strictures of journalism, if that’s what you’re into—is to hire writers and editors to produce it. We’re extremely available for work: per Axios, by mid-June 2023, more journalists had already been laid off (~1,900) than all of last year. Though Volv is currently hiring for several jobs, the closest thing to a writing role is “social media strategist.” They’re also seeking a “growth hacker” to “brainstorm and execute out-of-the-box experiments to target, and deliver marketing campaigns to drive organic growth.”
When I suggested to Vazirani that the best way to help writers—which she claims is core to Volv’s mission in a way I found sincere, but not persuasive—was to pay them salaries to do those jobs, she launched off on a tangent about how writers are actually getting hired based on their social audiences these days. (That’s broadly untrue my experience, but I’ve only been doing this 13 years, what do I know?) She believes her platform could be a conduit to that process. “We hope that independent writers are the ones with the power, and what we're building here is to facilitate that.” She also promised multiple times that if and when Volv begins making money (it’s currently running ads, though Vazirani claims it is not currently in the black), the company will introduce a revenue-sharing model to directly compensate writers for their performance on the app.
She seemed earnest on this point (as she did throughout our sometimes-contentious 40-minute interview.) I believe that she believes that. But this is a notoriously difficult nut to crack, and Volv or any company like it, must attract more investors to grow without revenue to even get to that point. Media’s moneymen (they’re mostly men, lol) have a dismal track record of equitably compensating workers even when they literally employ them. I tried to emphasize this to Vazirani by pointing out that even though she didn’t consider $150,000 a lot of money as a seed investment, it was still more money than I’ve ever earned in a year by a wide margin. Unproven moonshots reliant on janky AI get “incubated” by investors while proven journalists with decades of experience get laid off because the work has never been important enough to the people empowered to secure a sustainable future for the workers that produce it.
Two things can be true, and often are. Vazirani can genuinely want to build a platform that “empowers creators” (in the jargon), and Volv can be symptomatic of a problem that is poisoning media at its roots. Namely: in their quest to accrue ever-increasing returns on their capital, investors and the well-polished, pitch-ready founders that promise them “the future of media” have worsened innumerable lives, destroyed a decent middle-class livelihood, and eroded an entire nation of readers’ trust in the news in the process. Investment flooded into digital media last decade as newspapers withered on the vine because the former offered low overhead and primo returns. Many founders and investors got rich, even as they continued to pay their workers peanuts and fought bitter anti-union campaigns to keep it that way. Growth came from social media gimmicks and search-engine optimization until it didn’t. Then the bottom dropped out.
Today, the whole business is in shambles. Newspapers are still dropping like flies, but even “disruptive” digital publishers like BuzzFeed have begun to fail while bosses (who are extremely viable targets for automation, by the way!) and their patrons escape with fortunes they captured selling the pipe-dream of forever growth to workers who cared about the work and greater fools who absolutely didn’t. After a brief detour through the crypto casino, many of those same investors are now back at the media crapshoot to bet on AI-based garbage machines.
It’s not hard to see why. You don’t have to pay the machine shit, and offshore labor has no choice but to accept even fewer peanuts than its American counterpart to midwife the garbage onto the internet. That's the holy grail for the capital class and its allies, and always has been: capturing the value of labor without having to share it. The buzzy blather and the slick apps are just abstractions to make that alchemy more palatable and less obvious to unsuspecting readers, who can’t figure out why every website sucks now and anything worth reading is stuck behind a paywall. It’s a moral failure and a market failure all at once, prosecuted by the same interests as before: union-busters, rent-seekers, and snake-oil salesmen that care even less about journalists’ livelihoods than they do about journalism’s social value.
There are people making money here, and it’s not us, because it’s them. Same as it ever was. Call me old-fashioned, but I believe in a media future that isn’t built on the same mistakes of its past, and I’m ready to fight for it. What other choice do I have?
If you like booze (and ferocious commentary about the political economy of the booze industry), you can buy your discounted subscription to Fingers here.