WR 12: What’s old is new again (the year, AI, MLM, and crypto)
Weekly Reckoning for the week of 1/1/24 (that’ll take some getting used to)
Welcome back to the Ethical Reckoner, and happy new year! Tech news slowed but didn’t stop over the holidays, so in this Weekly Reckoning, we cover Substack’s Nazi problem, some AI-related court cases (one dramatic, one hilarious), and what may be the first AI actor on TV. Then, I muse on how crypto is basically MLM, but for dudes.
This slightly delayed WR is brought to you by: plumbing crises! wet-dry vacs! and (hopefully) non-toxic insulation!
The Reckonnaisance
The New York Times is suing OpenAI for copyright infringement
Nutshell: In a lawsuit filed right after I logged off for break, the Times alleges that because its articles were used to train ChatGPT, the bot can generate near-verbatim excerpts of paywalled articles (non-paywalled article about it available here).
More: Apparently the Times was in talks with OpenAI about coming to an “amicable resolution” (likely involving licensing fees or something of the sort, like the AP and Axel Springer are doing). But talks didn’t go well, and now the Times is asking that all of OpenAI’s models that trained on its articles (presumably including GPT-3.5 and GPT-4) be destroyed.1
Why you should care: As IP lawyer Cecilia Ziniti put it in an excellent thread arguing that this is the strongest generative AI copyright case yet: “What’s at stake? The future of AI innovation and the protection of creative content.” (Or they’ll come to an agreement like Google did with news publishers in Canada and we’ll never really know what the law actually requires.) But still: prestige newspaper vs hottest tech company of this decade? This’ll be a doozy.
Ugh, there are Nazis here, too.
Nutshell: Substack is coming under fire for allowing literal Nazis to publish and make money on its platform.
More: Of course, Substack takes 10% of subscription fees, so when Nazis profit, so does Substack. The co-founder of Substack published a note saying “we don’t like Nazis either” but that they won’t kick them off because they don’t believe “censorship” is the best way to combat hateful ideas. Rather, they believe that Substack is “not one platform, it is thousands of platforms” and thus you get to choose what to engage with (the same is true of the Internet, but web hosting services have still shut down Nazi sites). That doesn’t change the fact that Substack, the company, is actively profiting off of Nazi ideas: as Casey Newton recently wrote on Platformer, “The correct number of newsletters using Nazi symbols that you host and profit from on your platform is zero.” Platformer, one of Substack’s biggest publications, is threatening to leave the platform over the issue, with meetings scheduled for this week.
Why you should care: Well, this is a Substack publication, so we’re all involved in Substack in some way. Nobody pays for this, so no money is going to the people who are “not just OK with, but in principle supportive of, having loads of out-and-out Nazis on their platform.” And thus far, that seems to have been my line—I am still on Twitter, also a platform where leadership is cool with hateful content, even though I am no fan of Elon Musk or the people he allows on the site, but I would never pay for it and I block as many ads as I can. But I really don’t want to endorse people and platforms that are fine with making money off of Nazis. We’ll see what happens with the Platformer-Substack talks, and if my moral outrage can overcome my platform inertia. Watch this space. But if you, like my mother, are wondering if it’s ok to keep subscribing to non-Nazi publications on Substack, subscribing to free publications is probably roughly the moral equivalent of using Twitter, and subscribing to paid publications is somewhere between that and Twitter Blue (since 90% of the money is going to the author of the publication, rather than the platform owners).
Michael Cohen becomes the latest to submit nonexistent, AI-generated cases to court
Nutshell: Do👏Not👏Use👏Generative👏AI👏To👏Do👏Research👏Without👏Verifying👏Everything👏 (is what someone should have told Michael Cohen).
More: Cohen, Donald Trump’s felonious former “fixer” now asking for an early en to his supervised release, was apparently doing legal research with Google Bard, Google’s ChatGPT competitor, and sent his lawyer three citations that Bard made up. His lawyer did not check them and included them in a brief to the court. And this is after a highly-publicized case (also in Manhattan federal court) where other lawyers filed a brief with fake citations generated by ChatGPT and got fined $5000.
Why you should care: Lawyers keep using generative AI to write their briefs and being shocked that it makes up citations, despite the fact that it is literally just guessing what the most likely next chunk of text is for a given piece of text. While you may not be a lawyer or in need of one (may it stay that way), these cases indicate that people really don’t understand what generative AI is, especially when bundled with search engines—Cohen is quoted as saying that he thought Bard was “a supercharged search engine,” which it is not, but that’s kind of confusing when it’s bundled into a search engine. Still, generative AI is likely to have an impact on whatever your profession is, and it would behoove all of us to understand what it can and can’t do.
AI-generated actor makes her debut in China
Nutshell: On the heels of AI influencers comes Lili Ziren, an “AI actor” making her debut in the Chinese drama “I Am Nobody” [异人之下] (available on YouTube!).
More: Ziren plays Er Zhuang, a minor character described as “female humanoid who is a digital manifestation of a critically injured woman kept alive downstairs in the NDT Express warehouse” (ok, sure). She’s only in two episodes, but is apparently undetectable as AI.
Why you should care: When I first read this, I thought this could never happen in the US because the new SAG-AFTRA contract would prohibit it. But that’s actually not true! Apparently studios just have to notify the union if they intend to use synthetic performers and offer a “good faith” bargaining opportunity to allow humans to audition for the role. So, this may turn out to be a global proof of concept; although right now it’s likely far more expensive than using a human actor, public reception (at least in China) was generally positive.2
Extra Reckoning
I recently started listening to Season 1 of The Dream, which is about multi-level marketing (by recently, I mean yesterday, and yes I’m already five episodes in). One of the most interesting things it’s talked about so far is how MLM—where companies rely on individuals to sell the products to others and recruit them into becoming sellers as well—began as “women’s work” in the late 1800s and early 1900s (Avon was founded in 1886!). This was for a few reasons: companies wanted to sell “women’s products” like perfume and make-up, traveling salesmen were mistrusted, and women had access to other women and their homes. Women often had to work to support their families, but could only do so in the “informal economy” (WWII excepting) and still had to perform their wifely duties, so MLM slotted into a perfect niche. It also served as a social outlet for housewives, who could hold and attend Tupperware parties and group meetings. Of course, a lot of these companies are essentially pyramid schemes, and on average, 99.6% of participants lose money.
Why are we talking about MLM? Well, I recently read Zeke Faux’s Number Go Up about the rise and fall of crypto, and it got me thinking that cryptocurrency is basically MLM for dudes—including the pyramid schemes.3 I’m thinking about Celsius, which was a crypto lender that promised astronomical returns that were super-duper-safe. (Narrator: it was not super-duper-safe.) It was basically a Ponzi scheme, Celsius crashed, and crypto Ponzi schemes cost victims $7.8 billion in 2022.
Crypto scams use a lot of the same marketing techniques as MLM companies, but geared towards dudes. Instead of presenting it as ways to make a little money on the side and be part of a great community of women, it’s presented as a “brave” investment by manly actors and influencers. Celsius’s marketing also included a lot of “fight the man” rhetoric (the banks are evil, you’re being taken advantage of, etc etc) that creates a community of believers similar to traditional MLM, but considerably more sinister than a Tupperware party. And it encourages bringing other people in, not through trying to sell them a potentially useful product, but through FOMO4—as one NYT headline put it, “Everyone is Getting Hilariously Rich and You’re Not.”
Money and the idea that you are deserving of success is the foundation of all of this—MLM has roots in the “human potential movement” that emphasized developing human potential and reaping the financial rewards. It’s almost a secular version of prosperity theology, which holds that material success is a sign of divine favor. And there’s an intersection with crypto scams, too—the $4 billion OneCoin pyramid scheme was spread partly through churches in Samoa and Uganda, where they preached prosperity theology with a crypto twist. Crypto communities often have religious undertones—such as an unwavering belief that “number go up”—but are also marketed as ways to prove or improve yourself, to become an alpha male with the money that you need to earn to be a man. It’s not a coincidence that crypto scams have been promoted by macho influencers like Tom Brady and Jake Paul, or that FTX was in part brought down by a Twitter ego match between two male CEOs.
An entire thesis could be written about the gender dynamics of crypto vs MLM marketing, and I’m not about to turn this into one. But if you’re wondering where this leaves us, I guess the answer is positive: if we recognize that many crypto scams use the same techniques as MLM, that gives us a toolbox to use to combat them. Men love taking over women’s work (see: programming), but in the case of crypto, it’s coming back to bite them.
I Reckon…
that it’s maybe a bit odd that my two main reading genres of late are “crypto goes bust” and “feminist Greek mythology retellings.”
P.S. Americans in London: come hear me speak on a Democrats Abroad UK panel on AI and Elections at The Reform Club next month! RSVP here.
This is something that machine unlearning might come in handy for.
One interesting comment was that she can “never have her reputation tanked by a scandal,” which
Caveat: not all crypto is a scam, not all MLM is a pyramid scheme, and not all pyramid schemes are Ponzi schemes.
“Fear of missing out”
Thumbnail generated by DALL-E 3 via ChatGPT with the request “Generate an abstract painting in the style of DALL-E 2 that abstractly embodies the phrase "what's old is new again" with a consistent color palate”.