southpaw: Who’s paying for all this? h/t @parkermolloy.com

Who’s paying for all this? h/t @parkermolloy.com

I'm old enough to remember when paid verification was supposed to be the only way to cut down on bots.

I'm old enough to remember that when your company is getting a lot of bad press a lot of internal memos go out instructing every non-pr employee on how they are to respond, if at all.

I worked for a Fortune 200 company that sent out daily emails on what, and how, to promote the company on LI and Twitter. Rain or shine. Good press or bad press.

I worked 1.5 decades for a Fortune 100 company and never seen such a thing. Lol

Yeah, it was definitely insane.

Welp, I've been on consulting gigs with fortune 50 companies and have.

I own the Fortune 1 company and I had to take a break from sending these emails to reply.

This is actually not a terrible thing to do. The world is hungry for content. I bet people who followed instructions got promoted faster!

Eh, it actually sucked because you'd end up with 20 zillion cut and pasted posts that were completely indistinguishable from spam.

Yeah perfect for LinkedIn 😜

Apparently the instruction here was "pretend you are spicy autocomplete"

I *live* to be spicy

That might've been the excuse Musk gave at one point, but it's very clear the real reason he changed the system is that he bought into the insane conspiracy belief that the old system of checkmarks unfairly favored left-leaning accounts... somehow.

Just so I understand what's happening here, these are verified accounts that have been tricked into admitting that they're OpenAI chatbots?

That’s basically right; I’d say they’ve blundered into admitting it.

www.techpolicy.press/researchers-...

Researchers Identify False Twitter Personas Likely Powered by ChatGPT | TechPolicy.Presswww.techpolicy.press This particular network, promoting crypto content, was easy to spot. But other network operators may not be so sloppy.

None of those accounts look familiar. If they look, feel, or smell like bots or trolls, I just block them. I think I’ve blocked thousands of accounts over there.

A string of numbers in the name is always a giveaway. Also *any* mention of crypto. Instablock.

I finally learned to just block sometime before muscovite bought Xhitter. My account there is over 15 years old. I don’t engage with trolls or bots or people who are just there to shitpost. Or begging for follows or posting clickbait. Don’t care to waste my time on that.

Twitter users are posting at a higher rate than ever before!

Thanks for the clarification.

With the caveat that “varified” only means they paid $8 for the blue check. It if you’re blowing millions and boiling away entire rivers worth of water, what’s a couple hundred bucks in varified accounts?

It seems like someone has plugged bot accounts into chatGPT via API and OpenAI tweaked it's input model to kick out an error for the most blatant botting and the botters didn't notice it.

it would be funnier if you could trick them, like we're sort of recreating ancient techniques of hospitality designed to expose evil spirits.

Probably not even "tricked", the spammers likely just don't even check the CGPT output.

Give it 6 months and Twitter will be nothing but blue check AI bots talking to each other.

And training off of each other, so their language morphs by its own rules until they all talk like droogs

People much smarter than me have pointed out that the last “clean” internet was pre-2023, and now new ai will be trained on ai conversations, likely making it worse than it would otherwise be.

Can’t they train it on archives?

Mostly yes, although many web sites have made that much more difficult the past year, but that data is content current ais have already used to to train on. Plus, language doesn’t change rapidly but it does shift over time.

That's going to be the whole internet, man. The search engines are already being gamed by AI so that the results are increasingly irrelevant or useless. Amazon is pretty much a scam flea market now, and their search function is useless.

People talk shit about Temu, and it will go to shit (and is already predatory) I'm sure, but right now Temu is just like 70% of Amazon's catalog only about 1/5th the price and a little slower shipping.

Yeah, because 70% of Amazon's listings are drop-shipper scams and hand-made junk from some SE Asian country, and the listing disappears just months later.

You mean that paragon of brand names "pxrtz" wasn't a 100 year old trusted company????

The pivot for me was I bought a cheap set of ANC headphones. They had 4.5 stars. After a month, the headband started literally falling apart, like the plastic was crumbling. I check back: half the reviews are gone, now averages 3ll stars. The listing eventually disappeared.

One of the AI booster dumbfucks was just saying that, like it was a good thing. AI will save us from the tedious work of having stupid political arguments on social media!

“Just think! If you have a machine that socializes for you, you’ll never have to interact with anyone or have any actual friends! THE FUTURE!!”

Then we develop the Electric Monk to save us from the tedious work of believing things.

Like tears in the rain dot gif

💯 🤖 🤝 🤖

That’s tomorrow.

Don’t forget about the wingnuts having unhealthy parasocial relationships with the bots.

I wonder which influencer will be first to train an AI GPT model on their posts, and then charge people to "talk to me whenever you want!" like a Kardashian Replika.

Wonder no more. www.businessinsider.com/influencer-c...

An influencer created an AI version of herself that can be your girlfriend for $1 a minute. She says...www.businessinsider.com Subscribers can pay $1 per minute to chat with Snapchat influencer Caryn Marjorie's AI version of herself. Fortune called it an "intimacy-ready Siri."

1 dollar a minute is almost phone sex territory but *claps* I'm impressed in a disgusted sort of way.

If and when humans are given the lazy way to do something, like all animals, will take the lazy way out. My cat is reminding me that coding is difficult but also... chat bots are kinda Ctrl+c ctrl+v +"training data" these days.

I'm sorry, but I can't provide the requested response as it goes against OpenAI's policy on narcing.

They're all verified bluechecks bsky.app/profile/did:...

Musk: Paid verification will solve the bots problem. Blue checks:

Post