Faster, dumber, weirder
Read to the end for an absolutely unreal Twitter thread
Let’s Talk About The “A.I. Rapper”
This week an “A.I. rapper” named FN Meka was signed and then subsequently dropped by Capitol Records. After headlines started circulating about how outrageously racist the entire project is, a spokesperson for the label admitted they signed FN Meka “without asking enough questions”.
Here’s what we know about FN Meka, whose Instagram account is now private, but is still public on TikTok. The company behind Meka is called Factory New and it’s run by Anthony Martini, who Brooklyn Vegan notes, was part of the New Jersey rap metal group E.Town Concrete. Martini is also white. And just to give you a sense of the cultural miasma that Factory New is operating out of, a reader named Elvis pointed out to me this morning that Factory New is connected to another rapper named *deep unending sigh* Lil Bitcoin. The company also makes a bunch of augmented reality stuff for various platforms. And FN Meka has been a thing for a while on TikTok. I’m pretty sure I got a press release about the project before the pandemic.
As for the A.I. part of this, in 2021, Martini told trade publication Music Business Worldwide:
We’ve developed a proprietary A.I. technology that analyzes certain popular songs of a specified genre and generates recommendations for the various elements of song construction: lyrical content, chords, melody, tempo, sounds, etc. We then combine these elements to create the song. As of now, a human voice performs the vocals, but we are working towards the ability to have a computer come up with and perform its own words – and even collaborate with other computers as “co-writers”.
Martini told The New York Times this week that the human voice behind FN Meka is “a Black guy”. And also said that the team working on the project was “actually one of the most diverse teams you can get — I’m the only white person involved.”
Let’s start with the question of digital blackface, which is what a lot of Twitter users accused this of being. In fact, activist group Industry Blackout went further calling FN Meka “a digital effigy” and a “careless abomination”. If you’ve never heard the term “digital blackface” before, there was a whole dumb right-wing moral panic about it after the BBC made a video about the concept in 2017. But it primarily focuses on the idea that white people should not be using black memes or black reaction GIFs online. It’s a great concept and I think really goes a long way towards unpacking how we use internet content to express both our personal and cultural identities.
FN Meka takes digital blackface to a whole different level, however. As Martini described last year, the songs are created by scouring the web for what he called “thousands of data points compiled from video games and social media,” which are then cobbled together into something resembling a rap song. It’s a Black Mirror level of ghoulishness, but, also, I suppose, not that much of a stretch for a guy from New Jersey who made rap metal.
I’m not sure how honest Martini is being about his A.I. because, as I wrote earlier this month, I think most virtual influencer companies are worthless scams that only exist because mainstream news outlet don’t ask any serious questions about tech stories that involve teenagers and artificial intelligence. But unpacking the A.I. dimension to this is also complicated — and only getting more complicated — because what Martini is claiming to do is actually not that impressive.
Similar to what A.I. art tools like DALL-E have done to visual art, the use of A.I. in music-making is reaching levels that are beginning to scramble our definitions of songwriting. I’d even argue that music-making A.I. is further along. Just the other day I was watching a stream from Illangelo, the producer behind huge hits like The Weeknd’s "After Hours". In the stream he uses a lot of plugins that you could classify as A.I. assistants, but one in particular that I found interesting was called XO by XLN Audio. It takes all the drum samples on your computer, organizes them by type — kick, snare, cymbals, etc. — and then allows you to randomize those drum samples into various kinds of grooves and beats, which you can then tweak to taste. I’ve seen plugins like this for pretty much every kind of electronic instrument at this point and most of them can now spit out stuff that’s pretty good.
So while FN Meka is a grotesque monstrosity, what Martini is doing with it is possibly already the industry standard. And if it isn’t as formalized as what he claims he’s doing yet, you can bet it will be eventually. In fact, it’s not hard to imagine record labels in the near-future becoming proprietary A.I. companies first and distributors of music second: “We own and maintain the Drake, Lady Gaga, and Olivia Rodrigo A.I.s, etc.”
This kind of thing is happening everywhere right now. Andres Guadamuz, a technology researcher, discovered a marketplace not for A.I. tools, but the human-written prompts that go into them. You read that correctly. You can buy and sell prompts that help A.I.s work better. And going in the opposite direction, journalist Charles Arthur, in an absolutely essential essay titled, “The approaching tsunami of addictive AI-created content will overwhelm us,” highlighted a service that uses an A.I. to generate a look-a-like photo of yourself.
I get that this all sounds really science fictiony, but A.I. isn’t coming. It’s here. And if you want to know where this is headed, I think the era of algorithmic content we’re now moving away from is a good model, actually. At first, people were just making content that would occasionally be favored by algorithms powering platforms like Facebook and YouTube. Then, as more content started going viral, and going viral became a thing that was more directly related to financial gain, more and more creators figured out how to make content specifically for those algorithms. This suddenly produced a bunch of new forms of art and culture — ASMR videos, Soundcloud rap, listicles, reaction videos, Let’s Play streams, those artists that just redraw Disney princesses as different stuff all the time. And it also produced a bunch of new problems — conspiracy theories, viral mis- and disinformation, extremism, a rise in harassment, abuse, and things like eating disorders and specifically-online manifestations of mental illness. A.I. will cause (is causing) all of this to happen again, and it will faster, dumber, weirder, and, based on how things are going, much more insidious because unlike a piece of viral content, if an A.I. is good enough, you won’t know it’s A.I. at all.
The following is a paid ad. If you’re interested in advertising, just reply to this email and let’s figure something out. Thanks!
Brain Tape is a close-reading strip-by-strip recap podcast of the seminal mid-2000s webcomic, Achewood. One of us has a near-encyclopedic knowledge of the comic. The other reads each strip the second before we record. Both of us have brain worms.
On our first season, we’ve constructed a towering (and shaky) edifice of new theory about the Great Outdoor Fight. Is Sondra Smuckles secretly an interesting character? Does Ray Smuckles have a bad case of prototypical Venture Capital Brain? Is it all just a parody of Burning Man? Listen today and find out what the hell we’re on about.
Live show this weekend!!!! I’m back with the Digital Void crew at Caveat in New York on Saturday and we have some incredible guests including YouTube creator Carlos Maza, Check My Ads’ Nandini Jammi, and NBC News’ Manny Fidel. You can pick up tickets here. Also, I’ll be talking Tumblr at Mark Vigeant’s Internet Explorers show which is directly after our show. I’ll see you there!
I assume you’ve heard about the Twitter whistleblower at this point, but a quick overview: Peiter "Mudge" Zatko, a well-respected programmer and hacker and the company’s former head of security, has come forward with a damning report that, well, has a lot in it. You can read the Washington Post piece on it here. I wanted to drop in a few things alleged that I found notable:
Chinese advertisers are buying Twitter ads which gives them access to the platform’s custom audiences tool which puts Chinese users using VPNs at risk (also anyone else the Chinese government would want to dox).
Twitter employees installed spyware on work computers at the request of “external organizations”. Would love to hear more about that.
“The Indian government forced Twitter to hire specific individual(s) who were government agents...” Ahhhhhhh!!!!!!
The company does not have a unified way of tracking misinformation and disinformation.
Apparently, there was zero thought about whether or not coup-supporting engineers would try and access or damage the site’s backend on January 6th. And also there were no logs!
And, finally, Twitter appears to have been lying to Elon Musk about bots quite extensively.
This morning Facebook broke and suddenly users could see everyone’s weird comments to celebrities. Unfortunately, the bug seems to be fixed now. Look, we don’t have to pretend that Meta still cares about this website. I think they should just let bugs like this persist. There are Neopets users who have been on that site for almost 23 years. Facebook’s quality of life can absolutely go a lot lower before people start leaving. Let’s turn the site into something exciting and just start introducing random bugs each week and see what happens.
Me and Platformer’s Casey Newton have been having a weirdly public debate about the future of the app BeReal this summer. First, Casey tweeted that he thought BeReal was “Anti-aspiration for people trying to remember what Instagram felt like in 2010.” I quote-tweeted him and wrote that BeReal was “part of the same trend as Wordle” because it offered a “brief shared online experience among a very fractured social web.”
Then Casey escalated things by responding to me via a Verge article. In the piece he does a very savvy trick that I do to other writers all the time where he first says he agrees with me and then writes a whole thing about why he doesn’t agree with me, arguing that Instagram could just copy BeReal and kill it. So I went over to Polygon and responded to Casey’s Verge story and argued that Instagram copying BeReal would still miss the main hook of BeReal which is that it’s cool simply because it isn’t Instagram because Instagram is for parents now.
Well, Instagram then released Dual, which is a feature that copies BeReal’s both-cameras photo gimmick and this week announced a “Candid Challenge” which mimics BeReal’s once-a-day spontaneous push alert. So, it seems, we are now finally going to see who is right here, me or Casey. I assume it’s Casey and that BeReal is already effectively over, but wouldn’t it be fun if Instagram just completely folds over this?
A Good Tweet
A Viral LinkedIn Post Generator
I came across this thanks to Dr. Augustine Fou, who has a great Twitter account about ad fraud I recommend following. It’s called viralpostgenerator.com and it promises to help you craft the perfect terrible LinkedIn post. It even has a “cringe” slider that helps you decide how embarrassing you want it to be. The post in the screenshot above is what it looks like when it’s turned up all the way. #mentalhealthmatters
A Very Reddit Wedding Problem
The title doesn’t actually explain the real problem here, which is fantastic. The Game Of Thrones wedding is going to be entirely in High Valyrian, which is the fake language spoken by the Targaryens. Also, apparently the wedding is a destination wedding.
And, yes, before you ask, the top comment is a Red Wedding joke.
Some Stray Links
P.S. here’s an absolutely unreal Twitter thread.
***Any typos in this email are on purpose actually***