Cargo cults for autocorrect

You can find audio versions of Garbage Day on every major podcasting app. If it’s not there, here’s an RSS feed.

AI Doomerism Is Just Marketing

Last week, a group of former OpenAI employees came forward with a very scary warning that AI companies have been acting secretly and irresponsibly and need to be regulated. Was this group of insiders ringing the alarm bell on rampant copyright infringement? No. What about the fact that racial biases are being encoded into these models? Nope. Misinformation and destruction of journalism? Don’t be stupid.

Rather than anything that currently matters or even exists, this group is actually very worried about “the loss of control of autonomous AI systems potentially resulting in human extinction,” as they wrote on their website. It makes for good headlines and is enough to earn them the token New York Times arms-crossed whistleblower photo, but let’s be real. This is just marketing. That doesn’t mean I don’t think these people believe what they’re saying. They definitely seem to. And that’s the problem.

If you’re a normal person, you might not know that two competing groups of weird nerds in Silicon Valley have been locked in a cringe philosophical battle over what both sides believe is the inevitable rise of an artificial super intelligence. I wrote about this a few times last year, but here’s the best TL;DR I can give you.

On one side there’s the Effective Altruists (EA). They believe that human happiness is an equation that can be solved with money. And many of these people have, over the last few years, come to the loose consensus that an AI-powered society is the best way to manage this happiness equation. But they also believe AI could be an extinction-level event if it were to fall into the wrong hands.

On the other side are the effective accelerationists (e/acc), who don’t really care much about human happiness. They also believe that AI has the power to change society and they also don’t really care how it changes, only that it does and that they helped build the AI that did it.

It’s your classic managers vs. radicals schism. And, yeah, it makes for some fun newspaper stories, but, of course, doesn’t actually matter. The only people who pretend to even half-care about any of this are guys who write newsletters about AI, and they’re only doing that because writing about AI is a cheaper cost per acquisition for capturing email addresses right now than buying Facebook ads.

I think this is all very silly. I also think this the logical conclusion of rich, isolated computer programmers having ketamine orgies with each other. But it does, unfortunately, underpin every debate you’re probably seeing about the future of AI. Silicon Valley’s elite believe in these ideas so devoutly that Google is comfortable sacrificing its own business in pursuit of them. Even though EA and e/acc are effectively just competing cargo cults for a fancy autocorrect. Though, they also help alleviate some of the intense pressure huge tech companies are under to stay afloat in the AI arms race. Here’s how it works.

Generative AI generally moves in a boom and bust cycle. This is because both the technology it runs on and the attention spans of the humans that are meant to use it plateau very quickly. You get a new model, you get bored of it (because of the act of creation is actually the only interesting thing about creation and these tools remove that part for you), you forget about it, repeat. And we actually just came out of one of these plateaus.

As Big Technology’s Alex Kantrowitz argued back in February, citing Similarweb data, ChatGPT usage had actually been stagnating since May 2023. Last month’s release of ChatGPT-4o, a new free version of the service, helped ChatGPT shoot back to the top of the internet again, per Similarweb. But releasing a free version of your product doesn’t fix the main issue that everyone working in AI right now is trying to navigate: This stuff is expensive.

Analysts told The Information last year that OpenAI’s ChatGPT is possibly costing the company up to $700,000 a day to operate. Sure, Microsoft invested $13 billion in the company and, as of February, OpenAI was reportedly projecting $2 billion in revenue, but it’s not just about maintaining what you’ve built. The weird nerds I mentioned above have all decided that the finish line here is “artificial general intelligence,” or AGI, a sentient AI model. Which is actually very funny because now every major tech company has to burn all of their money — and their reputations — indefinitely, as they compete to build something that is, in my opinion, likely impossible (don’t @ me). This has largely manifested as a monthly drum beat of new AI products no one wants rolling out with increased desperation. But you know what’s cheaper than churning out new models? “Scaring” investors.

This is why OpenAI lets CEO Sam Altman walk out on stages every few weeks and tell everyone that its product will soon destroy the economy forever. Because every manager and executive in America hears that and thinks, “well, everyone will lose their jobs but me,” and continues paying for their ChatGPT subscription. As my friend Katie Notopoulos wrote in Business Insider last week, it’s likely this is the majority of what Altman’s role is at OpenAI. Doomer in chief.

This is also why all the AI whistleblowers that are crawling out of the woodwork right now aren’t advocating for just… no AI. And the best example of this is Leopold Aschenbrenner, another former OpenAI employee, and, as writer Max Read concisely put it, Gen Z “ambition psycho”. He’s shrewdly figured out how climb the ranks of divorced dad podcasts by applying Bitcoin maximalists’ obsession with white papers and misleading charts towards baiting AI doomers.

I’ve written this before, but I’m going to keep repeating it until the god computer sends me to cyber hell: The “two” “sides” of the AI “debate“ are not real. They both result in the same outcome — an entire world run by automations owned by the ultra-wealthy. Which is why the most important question right now is not, “how safe is this AI model?” It’s, “do we need even need it?”

The following is NOT a paid ad. It’s a good ol’ fashioned promo4promo. If you’re interested in promo swaps OR paid advertising, email me at [email protected] and let’s talk. Thanks!

Forget scrolling through endless feeds to catch the latest in social media. 

Matt Navarra does the heavy lifting with Geekout — delivering an information-packed newsletter every Friday. 

It's where 30,000+ social media and marketing professionals get their latest social media news, expert analysis, and an occasional meme, too.  

Join them today — it's free! 

A Good Cooking TikTok

@nordycos

A day in the life of Kaiba’s personal chef?! Kaiba: Me Chef: ironkid2.0 (not on tiktok) . . . #privatechef #hamptons #dayinthelife #yugioh... See more

How Does MrBeast’s Charity Work?

Last week, I published a bit of a scoop in Fast Company. MrBeast’s charity, Beast Philanthropy, received money from Stand Together, an organization founded by Charles Koch, which was described to me as the “umbrella organization for the Koch network.”

You can read the whole story here.

Cara Got Too Popular Too Fast

Interestingly enough, I’ve seen the most amount of chatter about Cara on Threads. Though, I think that might be a side effect of the fact that, best as I can tell, the biggest community on Threads are aggrieved Instagram creators that finally have a non-image-based platform to complain about Meta.

There is nothing particularly special about Cara, as a platform, it’s your standard art-hosting site. But it’s blowing up right now because it has adopted a very strong anti-AI policy. I’ve also seen some Instagram artists say they’re moving to Cara to get away from the platform mechanics that decide how art is shared elsewhere.

Though Cara’s new-found popularity has come with some downsides. First, their server bill is, reportedly, through the roof. Second, the site’s new users, who both hate AI and also don’t seem to fully understand how AI works and definitely don’t understand that AI services are not a monolith, started spreading rumors that Cara was lying about being anti-AI. This all stems from the fact that Cara uses a third-party service called Hive, which provides AI-based moderation for spam.

If you still don’t get how AI doomerism is a useful marketing strategy for power-hungry corporations that want to destabilize and then ultimately replace the entire internet by making it too shitty and chaotic to use, please read today’s top essay and and then read this section again and just repeat that over and over until it clicks for you.

The Plain White T’s Heard “Wan Gwan Delilah”

I’m still split on whether or not this song is supposed to be bad on purpose, but I’ve decided that, as a treat for myself, I’m just going to ignore that and continue to enjoy how horrible this song is, regardless.

Chappell Roan Popular

Chappell Roan had a massive crowd at Governor’s Ball over the weekend. Which has, predictably, led to a bunch of accusations that she’s an industry plant. This is primarily due to the fact that pop music stans have become feral now that there isn’t a digital media ecosystem aggregating their posts and giving them an illusion that they’re in charge of pop culture.

But just in case there are any questions about Roan, I thought NBC News’ Kat Tenbarge summed things up nicely, writing, “Chappell Roan does not have the money to pay Spotify to make her popular she literally lived with roommates as of last year.” And Fluxblog writer Matthew Perpetua added to that, writing, “The alternative is admitting they just didn't know about something coming up over the past few years and they need to hang on to a feeling of false superiority.” If you don’t know who Roan is, you will very soon. Her Tiny Desk set is a good entry point.

I don’t think it’s a coincidence that Roan is pulling in massive festival crowds the same weekend Charli XCX’s brat is racing towards a big Billboard debut. The vibe is finally shifting, folks.

It’s All Kicking Off In The SNL Slashfic Community

Alrighty, it’s been a while since we had a good and messy bit of fandom drama. Let’s dig in, shall we?

Last week, a writer named Chayla Wolfberg did a “you said normal girls were boring,” TikTok slideshow. Except, in her slideshow, she revealed that the weird thing that she did was self-publish romantic fanfic about Michael Che and Colin Jost from Saturday Night Live’s Weekend Update.

Unfortunately, for Wolfberg, users quickly noticed that on the cover, the character that was supposed to be Che in her story was portrayed as a white woman. That also kind of looked like her.

And so Wolfberg, first, posted a video explaining that she wrote Che’s character as a white woman because she didn’t feel like she should “take up space” writing a book featuring a queer, interracial relationship. She then seems to have deleted that, but there are screenshots here. Then she posted a video saying that her book is not actually SNL fanfic at all and that it’s just loosely based on that.

Alright! I’m glad that’s all sorted.

This Is The Vibe This Summer

@blond.in.car

Replying to @lifeofatrillionaire its charli babyyyy #360 #charlixcx #camp #omnichord #coversong #blondincar #autotune #glitter #girlypop #... See more

Some Stray Links

• “Jack Dorsey gave $10 million to a project run by an anonymous dev — who turned out to be a follower of a fascist 'guru'“

• “I Used to Run a Popular Newsletter. Then Things Started Getting Weird.” (I’m quoted in this!)

• Adam’s in WIRED writing about Skibidi Toilets!

P.S. here’s a good Spotify playlist.

***Any typos in this email are on purpose actually***